Gracefully handle failure to obtain writeable database

When we update the Xapian database, we grab a write lock on it. If two processes try to update the database at the same time, only one of them is able to get the lock, the other will fail to make any updates.

This patch detects this type of error, and responds accordingly. If immediate indexing is enabled but we fail to grab the write lock, we queue up the node to be indexed on the next cron run.