Add SaveLayer-Draw-Restore optimization.

Add SaveLayer-Draw-Restore optimization.

This is like SkPictureRecord's remove_save_layer1 but works with all draw calls.

Interesting patterns removed:
  SaveLayer-DrawRect-Restore: Silk SKPs, desk_weather
  SaveLayer-DrawPath-Restore: desk_carsvg, desk_wowwiki, tabl_androidpolice
  SaveLayer-DrawPosTextH-Restore: tabl_android_police
There may be others, but I stopped looking.


BUG=skia:2378
Committed: http://code.google.com/p/skia/source/detail?r=14610

[robertphillips, 2014-05-07 14:28:15]

lgtm

https://codereview.chromium.org/269813010/diff/60001/src/record/SkRecordOpts.cpp
File src/record/SkRecordOpts.cpp (right):

https://codereview.chromium.org/269813010/diff/60001/src/record/SkRecordOpts....
src/record/SkRecordOpts.cpp:99: const uint32_t  drawColor = 
drawPaint->getColor();
In some cases we could (in the distant future) combine the layer's alpha and the
draw paint's alpha.

https://codereview.chromium.org/269813010/diff/60001/tests/RecordOptsTest.cpp
File tests/RecordOptsTest.cpp (right):

https://codereview.chromium.org/269813010/diff/60001/tests/RecordOptsTest.cpp...
tests/RecordOptsTest.cpp:170: 
Not for this CL but, at some point we should render before and after tiles and
compare the colors.

[mtklein, 2014-05-07 14:38:14]

https://codereview.chromium.org/269813010/diff/60001/src/record/SkRecordOpts.cpp
File src/record/SkRecordOpts.cpp (right):

https://codereview.chromium.org/269813010/diff/60001/src/record/SkRecordOpts....
src/record/SkRecordOpts.cpp:99: const uint32_t  drawColor = 
drawPaint->getColor();
On 2014/05/07 14:28:15, robertphillips wrote:
> In some cases we could (in the distant future) combine the layer's alpha and
the
> draw paint's alpha.

Oh, right.  Noted.  Seems everything I could find is pure-alpha layer color and
pure-opaque paint color, but it'd be nice to be as general as possible.

https://codereview.chromium.org/269813010/diff/60001/tests/RecordOptsTest.cpp
File tests/RecordOptsTest.cpp (right):

https://codereview.chromium.org/269813010/diff/60001/tests/RecordOptsTest.cpp...
tests/RecordOptsTest.cpp:170: 
On 2014/05/07 14:28:15, robertphillips wrote:
> Not for this CL but, at some point we should render before and after tiles and
> compare the colors.

Yup, I think we're already covered.  We run SkRecord with and without
optimization in DM, comparing that both are identical to the direct raster
bitmap (and thus, to each other).  This pattern seems to be tested pretty
thoroughly by gm/optimizations, so we get all those before/after tests for free
here too.

I'm planning to also add a SKP-as-GM mode to DM so we can just toss in tricky
SKPs as an quick way to extend the tested set.

[mtklein, 2014-05-07 14:38:25]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-05-07 14:39:07]

CQ is trying da patch. Follow status at
 https://skia-tree-status.appspot.com/cq/mtklein@chromium.org/269813010/80001

[commit-bot: I haz the power, 2014-05-07 14:47:57]

Change committed as 14610