Throttling Breaks when Deleting Processing Jobs

I recently found an issue where jobs go straight to scheduled after using the delete feature on the dashboard. The job in question is using a [semaphore ](https://docs.hangfire.io/en/latest/background-processing/throttling.html#semaphores) to limit concurrent execution to 5 at a time. After deleting 5 jobs that were in the processing state all new jobs go straight to the scheduled state. I looked at the hangfire redis db and found what appears to be the semaphore for my job and once removing the 5 entries shown in the attached photo jobs began processing again. It appears deleting a processing job does not properly release the semaphore.


Start with the dashboard flow for deleting processing jobs and inspect how the semaphore state is stored in the Redis database. Reproduce the case by deleting five semaphore-limited processing jobs, then verify that the permits are released and subsequent jobs leave the scheduled state and process normally.

Written by the indexing model from the issue text.