Make Easy Unlock app have separate instance in incognito

Make Easy Unlock app have separate instance in incognito

Extension background page is only allowed for the incognito profile
on sign-in screen, so in order to run the app on sign-in, the app
must be run in split incognito mode.

To avoid having two background pages per user (for original and otr
profile) in regular user session, set split incognito mode only for the
app running on signin screen (by using different manifest than for a
regular session)

TEST=
1. Make sure easy sign-in works
2. Lock screen with an incognito window open.
   Verify that the screen can be unlocked using Smart Lock
BUG=448634, 445048
Committed: https://crrev.com/5a5bf0240c72ee259f29384f1fdf307a962a3ea2
Cr-Commit-Position: refs/heads/master@{#311543}

[tbarzic, 2015-01-14 03:35:16]

tbarzic@chromium.org changed reviewers:
+ xiyuan@chromium.org

[xiyuan, 2015-01-14 03:55:51]

Thank you for the super fast CL. :)

We might want to include 445048 with this CL as well. In case the Dmitry's CL is
merged, we need to merge this fix as well.

I was thinking of something like this when I file the bug. But I am a bit
hesitant now because we could have two background page running when inside a
user session with incognito:split. Would there be unwanted side effects (e.g.
double enroll/sync)?

[xiyuan, 2015-01-14 04:18:41]

I wonder if we should use two manifest, one for login screen and one for user
session.

[tbarzic, 2015-01-14 04:23:10]

On 2015/01/14 04:18:41, xiyuan wrote:
> I wonder if we should use two manifest, one for login screen and one for user
> session.

yeah, I think something like that should work..

[Dmitry Polukhin, 2015-01-14 07:18:58]

On 2015/01/14 04:23:10, tbarzic wrote:
> On 2015/01/14 04:18:41, xiyuan wrote:
> > I wonder if we should use two manifest, one for login screen and one for
user
> > session.
> 
> yeah, I think something like that should work..

We already use 2 manifests approach once for TTS. Perhaps we can use it here
too. I'm not going to merge my CL to branch because it might have complex side
effects like this CL :( I intentionally committed it after branch point to have
enough time for resolving consequences. Sorry for trouble that my CL caused.

[tbarzic, 2015-01-14 15:13:07]

OK, I split the manifests for apps running in user and sign-in profile. Along
the way removed some entries not needed by sign-in (we could probably remove few
permissions as well)
PTAL

[xiyuan, 2015-01-14 17:05:29]

Cool. LGTM

[tbarzic, 2015-01-14 17:06:27]

tbarzic@chromium.org changed reviewers:
+ sky@chromium.org

[tbarzic, 2015-01-14 17:06:28]

+sky for OWNER review (chrome/browser/browser_resources)

[sky, 2015-01-14 20:47:13]

.grd LGTM

[tbarzic, 2015-01-14 20:48:19]

The CQ bit was checked by tbarzic@chromium.org

[commit-bot: I haz the power, 2015-01-14 20:49:48]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/834233005/60001

[commit-bot: I haz the power, 2015-01-14 20:54:14]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2015-01-14 20:55:18]

Patchset 4 (id:??) landed as
https://crrev.com/5a5bf0240c72ee259f29384f1fdf307a962a3ea2
Cr-Commit-Position: refs/heads/master@{#311543}