[Crunch] crunch-job should not retry tasks after its SLURM allocation is revoked
|Velocity based estimate||-|
When a job is assigned to a single node, and it fails, SLURM revokes the job allocation. See #4410 for example logs. crunch-job correctly detects the node failure, but it then goes to retry the task, which is pointless without the job allocation. The job should fail immediately once the allocation is revoked.