Second attempt at the new syncer thread impl, now with less crashes!...

Second attempt at the new syncer thread impl, now with less crashes!
Previous one at http://codereview.chromium.org/214033.

I had local edits that resulted in initializing the CommandLine for syncapi,
but didn't have them as part of the patch, so this was causing a crash whenever
SyncerThreadFactory::Create was called.  The only diff here is the call to
CommandLine::Init in syncapi.cc.  This effectively means you can't change the
syncer thread impl on linux (we init an empty command line there), but this is
OK.  Once we link statically we won't need to do this.

TEST=ProfileSyncServiceTest, SyncerThreadTest, SyncerThreadWithSyncerTest


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=27281

[tim (not reviewing), 2009-09-25 17:05:56]

You'll notice syncer_thread.h/cc aren't in this Rietveld issue but they do show
up as part of the change locally as a server-side change only (a copy from
revision 27117).  My change somehow went through a black hole earlier and was
reverted, but SVN seems to be able to keep tabs on things even if they are in a
parallel universe.

[tim (not reviewing), 2009-09-25 18:39:32]

On 2009/09/25 17:05:56, timsteele wrote:
> You'll notice syncer_thread.h/cc aren't in this Rietveld issue but they do
show
> up as part of the change locally as a server-side change only (a copy from
> revision 27117).  My change somehow went through a black hole earlier and was
> reverted, but SVN seems to be able to keep tabs on things even if they are in
a
> parallel universe.

Yaayy everything is in the same universe again.  Backup-revert-overwrite
syncer_thread.{h,cc} seems to have done the trick.

[chron_chromium.org, 2009-09-25 20:16:59]

LGTM assuming nothing changed other than the one CommandLine::Init.