[BUG] Job cancelling fails at scale

[aaronleopold]

Describe the bug

A couple people have reported that attempting to cancel a scan when the library reaches a certain size fails. The description for what happens aligns between both reports:

• Attempt to cancel
• The thread seems to stay alive and updates are still coming through
• Attempting to cancel again results in a 500 because the JobManager already dropped the Worker

Killing the server fixes the issue

Expected behavior

Cancelling a job, regardless of how large the library is, should immediately kill the thread and stop execution

[aaronleopold]

Closing in favor of https://github.com/stumpapp/stump/issues/102