Fix FEImage SKP bounds

Fix FEImage SKP bounds

When recording its filterPicture, FEImage::createImageFilterForLayoutObject() transforms the
content such that it draws into dstRect - but it keeps the recording bounds at origin
(FloatRect(FloatPoint(), dstRect.size())).

This means that the picture content can draw outside its declared bounds, and is susceptible to
incorrect culling.

Since we're mapping the content into dstRect, we should use the same for picture culling.

R=senorblanco@chromium.org

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=200914

[f(malita), 2015-08-20 15:55:24]

This is not an issue currently, because picture bounds are only used for BBH
culling and the FEImage picture image filter is not part of that.

But it blocks https://codereview.chromium.org/1264133003/, which introduces more
aggressive picture culling at the SkCanvas level.  Layout test failures with
that patch applied, fixed by this CL:
https://storage.googleapis.com/chromium-layout-test-archives/linux_blink_rel/...

[Stephen White, 2015-08-20 16:00:42]

LGTM

[Stephen White, 2015-08-20 16:02:53]

BTW, not sure if it's relevant, but see
https://chromium.googlesource.com/skia/+/915881fe743f9a789037695f543bc6ea189c...
(which just landed).

[f(malita), 2015-08-20 16:08:28]

On 2015/08/20 16:02:53, Stephen White wrote:
> BTW, not sure if it's relevant, but see
>
https://chromium.googlesource.com/skia/+/915881fe743f9a789037695f543bc6ea189c...
> (which just landed).

Thanks.  Fortunately, none of that seems to depend on the SKP cull rect.

[f(malita), 2015-08-20 16:08:31]

The CQ bit was checked by fmalita@chromium.org

[commit-bot: I haz the power, 2015-08-20 16:08:54]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1298283003/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1298283003/1

[commit-bot: I haz the power, 2015-08-20 17:48:39]

Committed patchset #1 (id:1) as
https://src.chromium.org/viewvc/blink?view=rev&revision=200914