Prevent menus from blocking user activity detection

Prevent menus from blocking user activity detection

This is a retry CL for https://codereview.chromium.org/1024583003/ with
a slight change.

UserActivityDetector used to be an EventHandler triggered when the pre-
target event handlers are triggered. When a menu is present, the
MenuEventDispatcher doesn't trigger pre-target handlers and hence user
activity is not detected.
With this change the UserActivityDetector is made a PlatformEventObserver
to make sure it's always triggered.

R=sadrul@chromium.org, jamescook@chromium.org
BUG=462735
TEST=ui_base_unittests --gtest_filter=UserActivityDetectorTest.*

Committed: https://crrev.com/107bd9254664ba3c854c0d4fb3d2f1d85855fb16
Cr-Commit-Position: refs/heads/master@{#324366}

[afakhry, 2015-04-08 18:53:52]

@jamescook: Could you please review:
ash/shell.cc
extensions/shell/browser/shell_desktop_controller_aura.cc

@sadrul: Could you please review the remaining files?

The slight change here to the original CL is by having the UserActivityDetector
handle the received platform event in DidProcessEvent() instead of
WillProcessEvent(). Unless there is a bad side effect to this that I'm not aware
of, I think this will help us avoid any regression that may result from
converting a platform event to a ui::Event before actually dispatching the
events to handlers and doing this conversion again.

[James Cook, 2015-04-08 20:51:53]

ash/shell.cc and extensions/shell LGTM

[sadrul, 2015-04-09 01:13:08]

lgtm

On 2015/04/08 18:53:52, afakhry wrote:
> @jamescook: Could you please review:
> ash/shell.cc
> extensions/shell/browser/shell_desktop_controller_aura.cc
> 
> @sadrul: Could you please review the remaining files?
> 
> The slight change here to the original CL is by having the
UserActivityDetector
> handle the received platform event in DidProcessEvent() instead of
> WillProcessEvent(). Unless there is a bad side effect to this that I'm not
aware
> of, I think this will help us avoid any regression that may result from
> converting a platform event to a ui::Event before actually dispatching the
> events to handlers and doing this conversion again.

I actually don't think doing this in DidProcessEvent() makes a lot of difference
for this case. It would still trigger bugs in our repeat-event detection code.

[afakhry, 2015-04-09 01:16:10]

The CQ bit was checked by afakhry@chromium.org

[commit-bot: I haz the power, 2015-04-09 01:17:30]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1074543002/20001

[commit-bot: I haz the power, 2015-04-09 04:15:04]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-04-09 04:16:50]

Patchset 2 (id:??) landed as
https://crrev.com/107bd9254664ba3c854c0d4fb3d2f1d85855fb16
Cr-Commit-Position: refs/heads/master@{#324366}