make lock on republisher-checkin stronger

[danh]

It looks like in some rare cases, the lock on republisher-checkin is not strong enough to present races that start if two checkin processes are launched very near each other.

I think this could be fixed by the methods used in Bug #684492 for the scandata, or perhaps in this case we can do something simpler (like just checking for the existence of the file immediately before writing it ---- so we'd check existence twice: once at the start, and once when we start writing the lock file).

[danh (danh-archive)]

I should have noted that Paul referred me to a specimen in the wild that Hank caught, klinischestasche00corn from scribe3 in boston, uploaded 10/05/2011.

[danh (danh-archive)]

We added a second check for the existence of the lock file immediately before creating it (there was already a check, but much earlier). This is in revision 39141 (RePublisher-checkin.php).