Previously, when you did a large rejudging (say of a whole contest), abort it immediately, you would have a lot of queuetasks where each queue task doesn't have any remaining judge task to process. The judgedaemon would process these one at a time and assume there is no work to be done and sleep until processing the next one, so it would take hours to clean up the queue. We should address that behavior separately.
Previously, when you did a large rejudging (say of a whole contest), abort it immediately, you would have a lot of queuetasks where each queue task doesn't have any remaining judge task to process. The judgedaemon would process these one at a time and assume there is no work to be done and sleep until processing the next one, so it would take hours to clean up the queue. We should address that behavior separately.