Try to fix the memory leak reports in g_restrictions.

Try to fix the memory leak reports in g_restrictions.

This updates the Restrictions usage to be more consistent with other LazyInstances:
  - do not copy and assign
  - use static keyword
  - use LAZY_INSTANCE_INITIALIZER

BUG=106571
TEST=existing

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=117292

[Aaron Boodman, 2011-12-26 06:13:46]

I don't understand why this leaks. By default, LazyInstance is supposed to
delete itself at exit.

[jstritar, 2012-01-03 18:45:06]

Yeah, I think it should. 

The failures are occurring in UI tests, and there's a note at the bottom of the
"using valgrind" tests saying that UI tests aren't shutting down cleanly. Maybe
that's the issue?

[Aaron Boodman, 2012-01-03 19:26:58]

On 2012/01/03 18:45:06, jstritar wrote:
> Yeah, I think it should. 
> 
> The failures are occurring in UI tests, and there's a note at the bottom of
the
> "using valgrind" tests saying that UI tests aren't shutting down cleanly.
Maybe
> that's the issue?

OK, in that case I'd rather not add duct tape.

[jstritar, 2012-01-10 19:59:05]

Bringing back this one... 

If this was caused by the UI tests not exiting cleanly, then I'd expect more
LazyInstances would require suppressions.

I updated the usage to be more consistent with other LazyInstances. The
constructor and destructor is called roughly the same number of times, except in
the renderer. I think this is expected since renderers are killed.

I'll add the suppression back if the memory bots go red again.

[Aaron Boodman, 2012-01-10 23:42:10]

lgtm

[commit-bot: I haz the power, 2012-01-11 19:27:00]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jstritar@chromium.org/9027007/13001

[commit-bot: I haz the power, 2012-01-11 21:32:11]

Change committed as 117292

[jstritar, 2012-01-12 23:38:05]

FYI this stopped the Valgrind errors. Strange...