Timed out async queries making second callback

[tokebe]

It appears from internal discussion that BTE is failing to properly cancel effort on timed out tasks, causing a second callback to be sent to the client after the initial timeout.

This is possibly something wrong with either the redis queuing system, or more likely a failure in communication to the Piscina pool to stop the worker.

This should be testable by running a query that does finish with a timeout that is lower than its completion time. It's possible that this only affects callback rather than the actual job status, in which case debug breakpoints on the callback code will also be necessary.