cc: Add synthetic invalidations for dropped recordings

cc: Add synthetic invalidations for dropped recordings

Fix the case where a recording tile within a picture pile is dropped
(due to it being far away and partially invalidated) and tiles within
that recording tile (but not invalidated) are updated to point to a new
picture pile that is unable to raster them.

This patch adds synthetic invalidations that will drop tiles that no
longer have recordings that can raster them.  The downside of this is
that content far away might get dropped unnecessarily if there are
invalidations far away.

Alternate approaches that were considered and not implemented:

* Don't update the pile on tiles if the new pile can't raster the tile.
The downside of this is that on every activation you have to scale every
tile's content rect and then compare it against the pile's recorded
region to see if it is still contained, and then not setting the pile if
it isn't.

* Don't drop far away recordings on invalidation and instead maintain a
more complicated recorded region by subtracting out "unhandled"
invalidations.  This could make the recorded region much more
complicated than it already is, making creation of new tilings much more
expensive.

NOTRY=true
R=danakj@chromium.org
BUG=173567
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=184525

[danakj, 2013-02-25 23:39:52]

lgtm

[commit-bot: I haz the power, 2013-02-25 23:43:49]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/enne@chromium.org/12335079/1

[enne (OOO), 2013-02-26 00:18:13]

android_dbg_triggered_tests (the only place this is tested) runs cleanly, and it
builds on win, mac, and linux cleanly, so am landing this directly so it can
make a build for tonight.

[commit-bot: I haz the power, 2013-02-26 00:19:00]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/enne@chromium.org/12335079/1

[commit-bot: I haz the power, 2013-02-26 00:20:57]

Change committed as 184525