add localMatrix to gradient Descriptor

this consolidation will also help transisition to new flattening pattern, where we want to
have a flatten/unflatten method on all of the common gradient params (i.e. Descriptor).

BUG=skia:
R=egdaniel@google.com

Author: reed@google.com

Review URL: https://codereview.chromium.org/461643002
12 files changed