Improve error handling/diagnostics

[bedge]

Just had a case where manually aborting a job did not release the mutex.
Have not tracked down the specifics yet, but, regardless it happened.

Things I noticed were:

1. Not clear how to tell what job is the owner of the mutex
2. What's the appropriate action to take to release the lock? (Other than simply deleting the branch)

This does make me want:

1. Max wait timeout. In case an aborted job fails to release the mutex again, this would let the next in line caller force the issue and take ownership.
2. A notification channel for cases where the above happens