Roll Skia DEPS 3230->3244

Roll Skia DEPS 3230->3244

This will cause the following layout test to fail on all platforms:
GPU : fast/canvas/canvas-strokePath-shadow.html

After this change lands, I will update that test to be less fragile by
allowing for tolerance (at least +-1, but a 2 is probably better)
since the failing values had to be premultiplied and then unpremultiplied, which
varies a tiny bit between implementations (even between diff GPUs).

See trybot results in https://chromiumcodereview.appspot.com/9452031/
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=123479

[epoger, 2012-02-24 14:48:55]

[created my own CL from Mike's roll in
https://chromiumcodereview.appspot.com/9452031/ , so that I can land it]

[epoger, 2012-02-24 14:57:43]

[adding enne, who is current webkit gardener according to the schedule]

[reed1, 2012-02-24 15:08:44]

lgtm

[epoger, 2012-02-24 15:34:34]

The layout test fix will go here: 
https://bugs.webkit.org/show_bug.cgi?id=79488 ('Make
fast/canvas/canvas-strokePath-shadow.html allow for pixel tolerance')

[epoger, 2012-02-24 20:12:16]

On 2012/02/24 15:34:34, epoger wrote:
> The layout test fix will go here: 
> https://bugs.webkit.org/show_bug.cgi?id=79488 ('Make
> fast/canvas/canvas-strokePath-shadow.html allow for pixel tolerance')

The layout test fix has landed in WebKit as
http://trac.webkit.org/changeset/108835 .

Once that gets rolled into Chrome, the fast/canvas/canvas-strokePath-shadow.html
tests should stop failing in the chrome.webkit waterfall.