[ios] WebStateList owns all WebState it manages.

[ios] WebStateList owns all WebState it manages.

Flip the Tab and WebState ownership relationship so that WebState owns
(indirectly via the LegacyTabHelper tab helper) the Tab. This required:
-   Changing Contextual Seach and pre-renderer to own WebStates.
-   Adding observers and delegates for WebStateList events to create
    and cleanup the Tab when a WebState is added or detached from the
    WebStateList.

Change the WebStateList API to take WebState via std::unique_ptr<> to
enforce correct use of the API.

BUG=546222
Review-Url: https://codereview.chromium.org/2775623002
Cr-Commit-Position: refs/heads/master@{#461750}
Committed: https://chromium.googlesource.com/chromium/src/+/2c600c5850d71dc0afca1cd196698eddc2ab598d

[sdefresne, 2017-03-23 16:31:27]

sdefresne@chromium.org changed reviewers:
+ rohitrao@chromium.org

[sdefresne, 2017-03-23 16:31:28]

Please take a look.

[sdefresne, 2017-03-23 16:57:28]

The CQ bit was checked by sdefresne@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-03-23 16:57:48]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-03-23 17:29:49]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-03-23 17:29:50]

Dry run: Try jobs failed on following builders:
  ios-simulator on master.tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/ios-simulator/bui...)

[sdefresne, 2017-03-24 15:00:33]

The CQ bit was checked by sdefresne@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-03-24 15:00:58]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[sdefresne, 2017-03-24 15:11:05]

The CQ bit was checked by sdefresne@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-03-24 15:11:11]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[sdefresne, 2017-03-24 15:11:35]

Patchset #2 (id:20001) has been deleted

[sdefresne, 2017-03-24 15:15:08]

ping

[commit-bot: I haz the power, 2017-03-24 16:09:44]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-03-24 16:09:45]

Dry run: This issue passed the CQ dry run.

[rohitrao (ping after 24h), 2017-03-26 23:21:22]

Initial set of comments.  I need to take a closer look at some of these files.

Like we discussed last week, I think it might be helpful to make a quick
flowchart that shows which methods are involved when tabs are created, added to
the tabmodel, and closed/destroyed.

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/sess...
File ios/chrome/browser/sessions/tab_restore_service_delegate_impl_ios.mm
(right):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/sess...
ios/chrome/browser/sessions/tab_restore_service_delegate_impl_ios.mm:116: [model
closeTab:model.currentTab];
We still need to call the TabModel API for this, not WebStateList?

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
File ios/chrome/browser/tabs/legacy_tab_helper.h (right):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
ios/chrome/browser/tabs/legacy_tab_helper.h:32: // The Tab instance associated
with the WebState.
Should we document the lifetime of this Tab object?

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
File ios/chrome/browser/tabs/legacy_tab_helper.mm (right):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
ios/chrome/browser/tabs/legacy_tab_helper.mm:39: [tab_ onWebStateDestroyed];
Is this better than making Tab itself a WebStateObserver?

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
File ios/chrome/browser/tabs/tab_model.mm (right):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
ios/chrome/browser/tabs/tab_model.mm:138: web::SerializableUserDataManager*
user_data_manager =
Where did this move from?

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
File ios/chrome/browser/tabs/tab_model_closing_web_state_observer.mm (right):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
ios/chrome/browser/tabs/tab_model_closing_web_state_observer.mm:28: // If YES,
then the session is saved if the WebState is closed.
We only save sessions when we close the active tab?  What happens when you close
a background tab?

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
File ios/chrome/browser/tabs/tab_model_web_state_list_delegate.mm (right):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
ios/chrome/browser/tabs/tab_model_web_state_list_delegate.mm:23: void
TabModelWebStateListDelegate::WillAddWebState(web::WebState* web_state) {
FYI I think I'm going to need to call AttachTabHelpers() in the new app as well.
 I'm starting to hit the point where I'm adding code that assumes various
helpers exist.

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/ui/f...
File ios/chrome/browser/ui/fullscreen_controller.mm (left):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/ui/f...
ios/chrome/browser/ui/fullscreen_controller.mm:717: [webController
removeObserver:self];
When is this observer removed instead?

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/ui/o...
File ios/chrome/browser/ui/overscroll_actions/overscroll_actions_controller.mm
(left):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/ui/o...
ios/chrome/browser/ui/overscroll_actions/overscroll_actions_controller.mm:522:
[webController removeObserver:self];
Do we not need to remove webcontroller observers anymore?

[sdefresne, 2017-03-28 12:03:05]

Patchset #4 (id:80001) has been deleted

[sdefresne, 2017-03-28 15:15:05]

rohitrao: please take another look.

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/sess...
File ios/chrome/browser/sessions/tab_restore_service_delegate_impl_ios.mm
(right):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/sess...
ios/chrome/browser/sessions/tab_restore_service_delegate_impl_ios.mm:116: [model
closeTab:model.currentTab];
On 2017/03/26 23:21:21, rohitrao (OOO until 3-31) wrote:
> We still need to call the TabModel API for this, not WebStateList?

No, we could call WebStateList (-closeTab: is just a wrapper). The corresponding
code would be:

  WebStateList* web_state_list = [tab_model() webStateList];
  web_state_list->CloseWebStateAt(web_state_list->active_index());

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
File ios/chrome/browser/tabs/legacy_tab_helper.h (right):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
ios/chrome/browser/tabs/legacy_tab_helper.h:32: // The Tab instance associated
with the WebState.
On 2017/03/26 23:21:21, rohitrao (OOO until 3-31) wrote:
> Should we document the lifetime of this Tab object?

Done in ios/chrome/browser/tabs/tab_lifecycle.md (in the current CL, next
patchset).

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
File ios/chrome/browser/tabs/legacy_tab_helper.mm (right):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
ios/chrome/browser/tabs/legacy_tab_helper.mm:39: [tab_ onWebStateDestroyed];
On 2017/03/26 23:21:21, rohitrao (OOO until 3-31) wrote:
> Is this better than making Tab itself a WebStateObserver?

Tab is already a CRWWebstateObserver. I agree that only one of the two classes
need to be an observer (though I would have preferred the tab helper to be the
observer as most of the tab helpers are WebStateObserver in addition to being
WebStateUserData).

TL;DR: Done.

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
File ios/chrome/browser/tabs/tab_model.mm (right):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
ios/chrome/browser/tabs/tab_model.mm:138: web::SerializableUserDataManager*
user_data_manager =
On 2017/03/26 23:21:22, rohitrao (OOO until 3-31) wrote:
> Where did this move from?

From Tab -openerId & Tab -openerNavigationIndex. Note that in my followup CL
https://codereview.chromium.org/2771863002/ I move this to the free function
responsible for the serialisation of WebStateList. I cannot really leave the old
code in that CL as there is no correct location to initialise -openerId and
-openerNavigationIndex on Tab (it was done in the constructor because the opener
was known, but this is no longer the case after the current refactoring).

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
File ios/chrome/browser/tabs/tab_model_closing_web_state_observer.mm (right):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
ios/chrome/browser/tabs/tab_model_closing_web_state_observer.mm:28: // If YES,
then the session is saved if the WebState is closed.
On 2017/03/26 23:21:22, rohitrao (OOO until 3-31) wrote:
> We only save sessions when we close the active tab?  What happens when you
close
> a background tab?

We also save the session if the current Tab changes. See
TabModelSelectedTabObserver
-webStateList:didChangeActiveWebState:oldWebState:atIndex:userAction:. Since the
current Tab changes when the current Tab is closed this mean that we always save
the session when a Tab is closed. If we don't mind calling TabModel
-saveSessionImmediately: multiple time, then I can remove the
-webStateList:willDetachWebState:atIndex: as it is only there to allow the check
"atIndex == webStateList->active_index()".

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
File ios/chrome/browser/tabs/tab_model_web_state_list_delegate.mm (right):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/tabs...
ios/chrome/browser/tabs/tab_model_web_state_list_delegate.mm:23: void
TabModelWebStateListDelegate::WillAddWebState(web::WebState* web_state) {
On 2017/03/26 23:21:22, rohitrao (OOO until 3-31) wrote:
> FYI I think I'm going to need to call AttachTabHelpers() in the new app as
well.
>  I'm starting to hit the point where I'm adding code that assumes various
> helpers exist.

Ack.

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/ui/f...
File ios/chrome/browser/ui/fullscreen_controller.mm (left):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/ui/f...
ios/chrome/browser/ui/fullscreen_controller.mm:717: [webController
removeObserver:self];
On 2017/03/26 23:21:22, rohitrao (OOO until 3-31) wrote:
> When is this observer removed instead?

In Tab -onWebStateDestroyed:.

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/ui/o...
File ios/chrome/browser/ui/overscroll_actions/overscroll_actions_controller.mm
(left):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/ui/o...
ios/chrome/browser/ui/overscroll_actions/overscroll_actions_controller.mm:522:
[webController removeObserver:self];
On 2017/03/26 23:21:22, rohitrao (OOO until 3-31) wrote:
> Do we not need to remove webcontroller observers anymore?

This is done in Tab -onWebStateDestroyed:.

[Eugene But (OOO till 7-30), 2017-03-28 15:27:17]

eugenebut@chromium.org changed reviewers:
+ eugenebut@chromium.org

[Eugene But (OOO till 7-30), 2017-03-28 15:27:18]

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/ui/t...
File ios/chrome/browser/ui/tab_switcher/tab_switcher_controller.mm (right):

https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/ui/t...
ios/chrome/browser/ui/tab_switcher/tab_switcher_controller.mm:53: #import
"ios/web/web_state/ui/crw_web_controller.h"
Could you please remove this import and cleanup
ios/chrome/browser/ui/tab_switcher/DEPS

[sdefresne, 2017-03-28 15:33:04]

On 2017/03/28 15:27:18, Eugene But wrote:
>
https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/ui/t...
> File ios/chrome/browser/ui/tab_switcher/tab_switcher_controller.mm (right):
> 
>
https://codereview.chromium.org/2775623002/diff/60001/ios/chrome/browser/ui/t...
> ios/chrome/browser/ui/tab_switcher/tab_switcher_controller.mm:53: #import
> "ios/web/web_state/ui/crw_web_controller.h"
> Could you please remove this import and cleanup
> ios/chrome/browser/ui/tab_switcher/DEPS

Sure, done in patchset #5.

[rohitrao (ping after 24h), 2017-04-03 17:49:35]

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/tab...
File ios/chrome/browser/tabs/tab.mm (right):

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/tab...
ios/chrome/browser/tabs/tab.mm:1058: [self.webController
removeObserver:fullScreenController_];
Over IM, you mentioned that Tab is now a WebControllerObserver because you had
to call removeObserver before webStateDestroyed.  Don't we have the same problem
here (and for the overscrollActionsController_ below)?

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/tab...
File ios/chrome/browser/tabs/tab_model.mm (right):

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/tab...
ios/chrome/browser/tabs/tab_model.mm:242: // removeObserver: to be called first
otherwise a lot of unecessary work will
This removeObserver comment might be stale now?  What "unnecessary work" is
going to happen now?

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/tab...
ios/chrome/browser/tabs/tab_model.mm:246: // Unregister all observers after
closing all the tabs as some of them are
We used to do this before closing all tabs.  Do you remember why?  What could
potentially break now that we're unregistering observers after closure?

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/tab...
File ios/chrome/browser/tabs/tab_model_unittest.mm (left):

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/tab...
ios/chrome/browser/tabs/tab_model_unittest.mm:1067: TEST_F(TabModelTest,
SetParentModel) {
Should we rewrite/fix this test instead of deleting it?

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/ui/...
File ios/chrome/browser/ui/tab_switcher/tab_switcher_controller.mm (right):

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/ui/...
ios/chrome/browser/ui/tab_switcher/tab_switcher_controller.mm:834: Tab* tab =
[tabModel insertTabWithURL:URL
Could you use insertTabWithLoadParams here instead?

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/ui/...
ios/chrome/browser/ui/tab_switcher/tab_switcher_controller.mm:1201: [tabModel
closeTab:tab];
DCHECK_NE(NSNotFound, [tabModel indexOfTab:tab]), to make sure we got the right
tabmodel?

https://codereview.chromium.org/2775623002/diff/180001/ios/shared/chrome/brow...
File ios/shared/chrome/browser/tabs/web_state_list.h (right):

https://codereview.chromium.org/2775623002/diff/180001/ios/shared/chrome/brow...
ios/shared/chrome/browser/tabs/web_state_list.h:28: explicit
WebStateList(WebStateListDelegate* delegate);
This is potentially a change that could be split out into a separate CL.  (We
could switch all callers to use Owned in this CL, then remove all of the
Borrowed code in a followup.)

[sdefresne, 2017-04-04 13:47:57]

Please take another look.

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/tab...
File ios/chrome/browser/tabs/tab.mm (right):

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/tab...
ios/chrome/browser/tabs/tab.mm:1058: [self.webController
removeObserver:fullScreenController_];
On 2017/04/03 17:49:34, rohitrao wrote:
> Over IM, you mentioned that Tab is now a WebControllerObserver because you had
> to call removeObserver before webStateDestroyed.  Don't we have the same
problem
> here (and for the overscrollActionsController_ below)?

On IM I said that Tab need to be a WebControllerObserver because I have to call
-setDelegate: not -removeObserver: (i.e. CRWWebController DCHECK that delegate
is nil when -close is called but does not check that it has no observers).

So this is not a problem removing the observer at that point.

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/tab...
File ios/chrome/browser/tabs/tab_model.mm (right):

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/tab...
ios/chrome/browser/tabs/tab_model.mm:242: // removeObserver: to be called first
otherwise a lot of unecessary work will
On 2017/04/03 17:49:35, rohitrao wrote:
> This removeObserver comment might be stale now?  What "unnecessary work" is
> going to happen now?

The comment says that all observer registered with TabModel via -addObserver:
need to be cleared before the tabs are closed. That is, this comment repeat the
DCHECK([_observers empty]);

This comment is redundant, I agree, but no more stale than earlier. Moved it so
that it is attached to the DCHECK.

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/tab...
ios/chrome/browser/tabs/tab_model.mm:246: // Unregister all observers after
closing all the tabs as some of them are
On 2017/04/03 17:49:35, rohitrao wrote:
> We used to do this before closing all tabs.  Do you remember why?  What could
> potentially break now that we're unregistering observers after closure?

As _observers is empty (i.e. there are no TabModelObserver), the only observers
that are going to be called now are the WebStateList observer registered by
TabModel, that is the observer that contains code extracted from TabModel as
part of my refactoring to introduce WebStateList.

Clearing the list before closing all the tabs was a bug I introduced as it
caused some of the code extracted from TabModel during the refactoring to not be
called when closing the tabs. Before the refactoring that code was invoked, so
it is fixing a regression I introduced.

TL;DR: nothing should be broken by this change, instead it may fix regressions
(though, this only happens when the app is closed, so most likely no visible
regression).

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/tab...
File ios/chrome/browser/tabs/tab_model_unittest.mm (left):

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/tab...
ios/chrome/browser/tabs/tab_model_unittest.mm:1067: TEST_F(TabModelTest,
SetParentModel) {
On 2017/04/03 17:49:35, rohitrao wrote:
> Should we rewrite/fix this test instead of deleting it?

Done. Also added a test that check a Tab is created when a WebState is added to
TabModel's WebStateList.

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/ui/...
File ios/chrome/browser/ui/tab_switcher/tab_switcher_controller.mm (right):

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/ui/...
ios/chrome/browser/ui/tab_switcher/tab_switcher_controller.mm:834: Tab* tab =
[tabModel insertTabWithURL:URL
On 2017/04/03 17:49:35, rohitrao wrote:
> Could you use insertTabWithLoadParams here instead?

Done.

https://codereview.chromium.org/2775623002/diff/180001/ios/chrome/browser/ui/...
ios/chrome/browser/ui/tab_switcher/tab_switcher_controller.mm:1201: [tabModel
closeTab:tab];
On 2017/04/03 17:49:35, rohitrao wrote:
> DCHECK_NE(NSNotFound, [tabModel indexOfTab:tab]), to make sure we got the
right
> tabmodel?

Done.

https://codereview.chromium.org/2775623002/diff/180001/ios/shared/chrome/brow...
File ios/shared/chrome/browser/tabs/web_state_list.h (right):

https://codereview.chromium.org/2775623002/diff/180001/ios/shared/chrome/brow...
ios/shared/chrome/browser/tabs/web_state_list.h:28: explicit
WebStateList(WebStateListDelegate* delegate);
On 2017/04/03 17:49:35, rohitrao wrote:
> This is potentially a change that could be split out into a separate CL.  (We
> could switch all callers to use Owned in this CL, then remove all of the
> Borrowed code in a followup.)

When rebasing, I had conflict with code that added new uses of WebStateList (in
unit tests). As I'm changing the API and behaviour of WebStateList, I prefer to
do it in a single CL to avoid either:
1. supporting non-owning WebStateList with a CloseWebStateAt method (weird),
2. chasing DCHECK failures in tests that tries to create non-owning
WebStateList.

Unless you strongly prefer this separation, I'd like to not do it.

[rohitrao (ping after 24h), 2017-04-04 14:07:36]

lgtm

[sdefresne, 2017-04-04 14:25:28]

The CQ bit was checked by sdefresne@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-04-04 14:25:46]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-04-04 14:33:53]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-04-04 14:33:54]

Dry run: Try jobs failed on following builders:
  ios-simulator-xcode-clang on master.tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/ios-simulator-xco...)

[sdefresne, 2017-04-04 15:21:10]

The CQ bit was checked by sdefresne@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-04-04 15:21:46]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[sdefresne, 2017-04-04 16:25:18]

The CQ bit was unchecked by sdefresne@chromium.org

[sdefresne, 2017-04-04 16:25:23]

The CQ bit was checked by sdefresne@chromium.org

[sdefresne, 2017-04-04 16:25:24]

The patchset sent to the CQ was uploaded after l-g-t-m from
rohitrao@chromium.org
Link to the patchset: https://codereview.chromium.org/2775623002/#ps240001
(title: "Fix gn check")

[commit-bot: I haz the power, 2017-04-04 16:25:44]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-04-04 16:50:03]

CQ is committing da patch.

Bot data: {"patchset_id": 240001, "attempt_start_ts": 1491323123438620,
"parent_rev": "19e6e13cfa014abe86feba8dbb549190becb2304", "commit_rev":
"2c600c5850d71dc0afca1cd196698eddc2ab598d"}

[commit-bot: I haz the power, 2017-04-04 16:51:39]

Description was changed from

==========
[ios] WebStateList owns all WebState it manages.

Flip the Tab and WebState ownership relationship so that WebState owns
(indirectly via the LegacyTabHelper tab helper) the Tab. This required:
-   Changing Contextual Seach and pre-renderer to own WebStates.
-   Adding observers and delegates for WebStateList events to create
    and cleanup the Tab when a WebState is added or detached from the
    WebStateList.

Change the WebStateList API to take WebState via std::unique_ptr<> to
enforce correct use of the API.

BUG=546222
==========

to

==========
[ios] WebStateList owns all WebState it manages.

Flip the Tab and WebState ownership relationship so that WebState owns
(indirectly via the LegacyTabHelper tab helper) the Tab. This required:
-   Changing Contextual Seach and pre-renderer to own WebStates.
-   Adding observers and delegates for WebStateList events to create
    and cleanup the Tab when a WebState is added or detached from the
    WebStateList.

Change the WebStateList API to take WebState via std::unique_ptr<> to
enforce correct use of the API.

BUG=546222

Review-Url: https://codereview.chromium.org/2775623002
Cr-Commit-Position: refs/heads/master@{#461750}
Committed:
https://chromium.googlesource.com/chromium/src/+/2c600c5850d71dc0afca1cd19669...
==========

[commit-bot: I haz the power, 2017-04-04 16:51:40]

Committed patchset #11 (id:240001) as
https://chromium.googlesource.com/chromium/src/+/2c600c5850d71dc0afca1cd19669...