Add functions for inspecting SkShader.

Add functions for inspecting SkShader.

Add a function to inspect an SkShader to determine if it is an
SkComposeShader.

Add a virtual function for determining if an SkShader is a custom
shader, which returns a custom set of information. The
implementation is in Android, and this function is only defined
if SK_BUILD_FOR_ANDROID_FRAMEWORK.

BUG=b/10650594

Committed: http://code.google.com/p/skia/source/detail?r=14716

[djsollen, 2014-05-13 17:15:40]

rebased clone of https://codereview.chromium.org/272503011/. Since Leon is still
out and I can't push the original through the commit queue as it fails to rebase
I uploaded this CL.

[reed1, 2014-05-13 18:00:23]

lgtm w/ request for dox (now or later)

https://codereview.chromium.org/282733005/diff/1/include/core/SkShader.h
File include/core/SkShader.h (right):

https://codereview.chromium.org/282733005/diff/1/include/core/SkShader.h#newc...
include/core/SkShader.h:361: *  fill it out with info about the shader.
? What is the ownership-rule for the returned pointers? 

I presume the answer is that they are bare-pointers, and that if the caller
wants to retain them, he needs to call ref. But no matter what we should
document this (now or in a later CL).

[djsollen, 2014-05-13 18:06:57]

The CQ bit was checked by djsollen@google.com

[commit-bot: I haz the power, 2014-05-13 18:07:13]

CQ is trying da patch. Follow status at
 https://skia-tree-status.appspot.com/cq/djsollen@google.com/282733005/1

[djsollen, 2014-05-13 18:07:55]

I'll defer the commenting to Leon.  I'll still submit this so we can unblock the
merge of Skia into android.

[commit-bot: I haz the power, 2014-05-13 18:14:48]

Change committed as 14716

[scroggo, 2014-05-14 13:51:25]

https://codereview.chromium.org/282733005/diff/1/include/core/SkShader.h
File include/core/SkShader.h (right):

https://codereview.chromium.org/282733005/diff/1/include/core/SkShader.h#newc...
include/core/SkShader.h:361: *  fill it out with info about the shader.
On 2014/05/13 18:00:24, reed1 wrote:
> ? What is the ownership-rule for the returned pointers? 
> 
> I presume the answer is that they are bare-pointers, and that if the caller
> wants to retain them, he needs to call ref. But no matter what we should
> document this (now or in a later CL).

Uploaded https://codereview.chromium.org/288713002