GitLab

This test shuts down the session service part way through,
and then recreates it. The second service ends up with a
different task runner (because each session service has its
own). This means it's possible for the second one to try
reading files that haven't been finished writing. At least
that's my theory as to why this test is flaky.

I couldn't repro this locally, so this is a best guess.

BUG=1154345, 1158715, 1166756
TEST=test only change, SessionRestoreTest.NormalAndPopup

Change-Id: I7bf5ea96134179f607ffcf9d882b6bc7b2277da2
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2634013Reviewed-by: David Bienvenu <davidbienvenu@chromium.org>
Commit-Queue: Scott Violet <sky@chromium.org>
Cr-Commit-Position: refs/heads/master@{#844842}