Bench baseline for mostly 0 image.

Bench baseline for mostly 0 image.

Add a baseline for decoding a mostly 0 image. This is in advance of
https://codereview.chromium.org/24269006/ which provides an option to
skip writing those 0s (as part of BUG=skia:1661
). On my Nexus 4, the
benchmark does not slow down after that change.

As suggested in https://codereview.chromium.org/24269006/ add a
resourcePath flag to bench. Will require a change in buildbot in order
to actually use the flag.

Add an image used by the test.

Until https://codereview.chromium.org/24448002 is submitted,
the test will not actually be run by the bots (since it
won't know where to find the file).

BUG=skia:1661
R=djsollen@google.com, mtklein@google.com

Committed: https://code.google.com/p/skia/source/detail?r=11461

[djsollen, 2013-09-25 13:38:25]

lgtm

[mtklein, 2013-09-25 13:48:31]

lgtm

https://codereview.chromium.org/24440002/diff/8001/bench/SkipZeroesBench.cpp
File bench/SkipZeroesBench.cpp (right):

https://codereview.chromium.org/24440002/diff/8001/bench/SkipZeroesBench.cpp#...
bench/SkipZeroesBench.cpp:101: SkImageDecoder* fDecoder;
SkAutoTDelete?

[scroggo, 2013-09-25 17:54:57]

https://codereview.chromium.org/24440002/diff/8001/bench/SkipZeroesBench.cpp
File bench/SkipZeroesBench.cpp (right):

https://codereview.chromium.org/24440002/diff/8001/bench/SkipZeroesBench.cpp#...
bench/SkipZeroesBench.cpp:101: SkImageDecoder* fDecoder;
On 2013/09/25 13:48:32, mtklein wrote:
> SkAutoTDelete?

Done.

[scroggo, 2013-09-25 21:42:20]

Committed patchset #2 manually as r11461 (presubmit successful).