Revert of DatabaseBackend should not post a task to an already terminated database thread

Revert of DatabaseBackend should not post a task to an already terminated database thread (patchset #4 id:60001 of https://codereview.chromium.org/596083005/)

Reason for revert:
This causes a lot of database layout test failures on the buildbots. Reverting.

Original issue's description:
> DatabaseBackend should not post a task to an already terminated database thread
> 
> Currently some methods of DatabaseBackend post a task to a database thread without checking DatabaseThread::terminationRequested(). Consequently, the task can be posted to an already terminated database thread.
> 
> This is problematic but by accident isn't currently causing any real issue (for some reason). This begins to cause an issue when I destruct DatabaseThread::m_thread promptly after the database thread processed a cleanup task (See https://codereview.chromium.org/589363002).
> 
> This CL adds the necessary DatabaseThread::terminationRequested()s before DatabaseBackend posts a task to the database thread.
> 
> BUG=340522
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=182552

TBR=oilpan-reviews@chromium.org,tkent@chromium.org,haraken@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=340522
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=182564

[Mads Ager (chromium), 2014-09-24 06:42:16]

Created Revert of DatabaseBackend should not post a task to an already
terminated database thread

[commit-bot: I haz the power, 2014-09-24 06:43:30]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/593423002/1

[commit-bot: I haz the power, 2014-09-24 06:44:08]

Committed patchset #1 (id:1) as 182564

[haraken, 2014-09-24 07:04:06]

LGTM, thanks for the revert.

Though I doubly-confirmed that the CL passed the test (and it actually passed
the commit queue)... Let me investigate.