In times of high contention, you could get the same scheduled rule added to the queue multiple times before they get run. When this occur, the rule will cancel instances of the execution of itself until the last instance of the execution. This can result in a rule not running for longer than appropriate, even given the lower priority of the queue.
We should fix this up so that the schedule only gets added iff the previous rule execution has fully completed. We should also give the user better logging when the rule fails or a manual run occurs