Notification implementation is clunky because it does not change the state of jobs entering notification

The notification phase of crawling is gated by the Job Notification Thread, which does not actually change the state of the jobs undergoing notification. This is a potential future problem because there may someday be more than one thread doing it.