Fix --trace-shutdown with thread-local tracing.

Fix --trace-shutdown with thread-local tracing.

With thread-local tracing (r221766), TraceLog::Flush() is async and
needs to be called from a thread having a message loop. It also requires
that tracing has been stopped.

BrowserShutdownProfileDumper is a special caller that TraceLog::Flush()
was originally called from a thread without message loop.

Now start another thread for flushing, and call TraceLog::SetDisabled()
before TraceLog::Flush().

The calling thread needs to wait for the completion of the flush,
otherwise the browser may shutdown before all trace events written.
Temporarily allow the thread to wait in ThreadRestrictions.

BUG=none
TEST=manual run chrome with --trace-shutdown. Should not assert. Should generate correct trace file.

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=223819

[jam, 2013-09-16 16:42:14]

can you get someone familiar with tracing internals to review this? i can
rubberstamp after

[Xianzhu, 2013-09-16 16:51:19]

+nduca for trace_event.

[nduca, 2013-09-17 19:32:31]

lgtm

[jam, 2013-09-17 20:24:38]

lgtm

[commit-bot: I haz the power, 2013-09-17 23:14:12]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/wangxianzhu@chromium.org/24047007/1

[commit-bot: I haz the power, 2013-09-18 02:54:33]

Retried try job too often on win_rel for step(s) browser_tests,
chrome_frame_net_tests, chrome_frame_tests, chrome_frame_unittests,
content_browsertests, mini_installer_test, nacl_integration,
sync_integration_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...

[commit-bot: I haz the power, 2013-09-18 03:57:52]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/wangxianzhu@chromium.org/24047007/1

[commit-bot: I haz the power, 2013-09-18 08:15:49]

Change committed as 223819