GitLab

Currently, gesture nav affordance uses a single layer that paints
everything in each frame. This CL moves the arrow to its own layer which
is drawn once and then appropriate transform and opacity are applied on
it. This is needed for implementing pull-to-refresh which needs the
arrow be rotated. Also, generally we want to decompose the affordance
into a few layers to avoid painting it on each frame.

BUG=702404,743239
TEST=none

Change-Id: I9140471d6b6edbbb84f97fd5148994fea58e5a90
Reviewed-on: https://chromium-review.googlesource.com/594840Reviewed-by: Sadrul Chowdhury <sadrul@chromium.org>
Commit-Queue: Mohsen Izadi <mohsen@chromium.org>
Cr-Commit-Position: refs/heads/master@{#491261}