GitLab

Previously, there was a duplicate observer that was (apparently?) okay
because spinning the current message loop was enough to load th
history backend.

The TaskScheduler migration requires us to make a different call to
ensure that the history backend is loaded.

This CL also removes the member observer, because running the
TaskScheduler until idle should be sufficient to assert that the
history backend has been loaded.

Bug: 751031
Change-Id: Iecacecddc3a50be3d54be64fe520fad757aa0275
Reviewed-on: https://chromium-review.googlesource.com/825642Reviewed-by: Peter Kasting <pkasting@chromium.org>
Commit-Queue: Tommy Li <tommycli@chromium.org>
Cr-Commit-Position: refs/heads/master@{#524484}