Re-land: Enable GpuMemoryBuffers video frames on MacOS.

Re-land: Enable GpuMemoryBuffers video frames on MacOS.

This patch enables UYVY GpuMemoryBuffers backed VideoFrames on MacOSX
for software decoded videos.
This change allows VideoLayers to be promoted to overlays when possible.
Power measurements showed a significant reduction in Watts consumed
during video playback: in the tests ran by ccameron@ it went from 3.15
Watts to 2.13 Watts.

Additionally this change avoids one copy of the VideoFrame in the GPU
process and moves away another copy from the Compositor thread to
media.

Unfortunately, for videos where the decoding time is close to the frame
time, the additional copy might slow down the decoding past the limit,
causing more dropped frames in these cases.

More benchmark results can be found here:
https://x20web.corp.google.com/~dcastagna/results_all.html

The first run is with GMB video frames disabled, the second one, enabled.

BUG=485859, 510252
Committed: https://crrev.com/96463459619f441a721f96c51c357208604b7475
Cr-Commit-Position: refs/heads/master@{#345418}

Committed: https://crrev.com/b65e607913fe32132494e877bc1b70108e4d0e20
Cr-Commit-Position: refs/heads/master@{#347566}

[Daniele Castagna, 2015-08-21 20:51:56]

dcastagna@chromium.org changed reviewers:
+ andresantoso@chromium.org, ccameron@chromium.org

[ccameron, 2015-08-21 21:00:46]

This lgtm -- I think we should try to enable for M46 branch point.

[Andre, 2015-08-21 21:10:20]

LGTM

[Avi (use Gerrit), 2015-08-21 21:34:23]

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

[Avi (use Gerrit), 2015-08-21 21:34:24]

content lgtm

[Daniele Castagna, 2015-08-21 21:35:07]

+avi for content/

Thank you Avi! :)

[Daniele Castagna, 2015-08-21 21:44:38]

dcastagna@chromium.org changed reviewers:
+ piman@chromium.org

[Daniele Castagna, 2015-08-21 21:44:39]

+piman for chrome/browser/chromeos/login

[piman, 2015-08-21 23:50:25]

lgtm

[DaleCurtis, 2015-08-25 18:59:12]

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

[DaleCurtis, 2015-08-25 18:59:13]

lgtm

[Daniele Castagna, 2015-08-25 19:02:49]

The CQ bit was checked by dcastagna@chromium.org

[commit-bot: I haz the power, 2015-08-25 19:03:57]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1308913003/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1308913003/1

[commit-bot: I haz the power, 2015-08-25 20:10:44]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-08-25 20:11:25]

Patchset 1 (id:??) landed as
https://crrev.com/96463459619f441a721f96c51c357208604b7475
Cr-Commit-Position: refs/heads/master@{#345418}

[mnaganov (inactive), 2015-08-25 23:29:59]

A revert of this CL (patchset #1 id:1) has been created in
https://codereview.chromium.org/1313123008/ by mnaganov@chromium.org.

The reason for reverting is: Suspicious for breaking Mac OS Blink layout tests
that play videos.

BUG=524733
.

[Daniele Castagna, 2015-09-05 06:03:40]

On 2015/08/25 at 23:29:59, mnaganov wrote:
> A revert of this CL (patchset #1 id:1) has been created in
https://codereview.chromium.org/1313123008/ by mnaganov@chromium.org.
> 
> The reason for reverting is: Suspicious for breaking Mac OS Blink layout tests
that play videos.
> 
> BUG=524733
> .

Re-landing this patch after a new extension CHROMIUM_ycbcr_422_image
(crrev.com/1305153005, crrev.com/1308643003, crrev.com/1316993003) has been
added.
The new extension is checked (crrev.com/1306243010) to select the output format
that is going to be used when converting videoframes, in this way
PIXEL_FORMAT_UYVY will be used only if CHROMIUM_ycbcr_422_image is available.
Previously UYUV would always be used causing incorrect video rendering when not
available.

Additionally, VideoRendererImpl::pending_read_ logic has been fixed
(crrev.com/1319303005) since it was causing some test flakiness when async
videoframe copies were enabled.

[Daniele Castagna, 2015-09-05 06:03:49]

The CQ bit was checked by dcastagna@chromium.org

[commit-bot: I haz the power, 2015-09-05 06:03:54]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1308913003/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1308913003/1

[commit-bot: I haz the power, 2015-09-05 07:18:51]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-09-05 07:19:45]

Patchset 1 (id:??) landed as
https://crrev.com/b65e607913fe32132494e877bc1b70108e4d0e20
Cr-Commit-Position: refs/heads/master@{#347566}