Potential deadlock for a workspace when a queue/worker is missing

[Mehmed Mustafa]

Current behavior:
When a workflow contains a call to a processor that was not deployed through a processing worker, the processing server would still lock the workspace pages and warn that a queue is missing. After deploying a worker manually (and the queue indirectly through the worker) and reposting the workflow will end up being in a deadlock for that same workspace since the locked pages were never unlocked.

Desired behavior:
Either the processing server should not lock anything before it is certain that the queue/worker for that request exists, or should unlock after a warning for the missing queue.

This is slightly behind my priority list but still, something I would fix as soon as possible.