Avoid crashing in this case, just assume concurrency limit is 1. This shouldn't happen with normal operation but we've had this problem when jobs that are concurrency-limited are enqueued from new code and dispatched from old code, where the job still doesn't have concurrency controls, thus returning nil for the limit and crashing the dispatcher.
Avoid crashing in this case, just assume concurrency limit is 1. This shouldn't happen with normal operation but we've had this problem when jobs that are concurrency-limited are enqueued from new code and dispatched from old code, where the job still doesn't have concurrency controls, thus returning
nil
for the limit and crashing the dispatcher.