available_worker doesn't handle worker exit

[ElectronicRU]

With strategy available_worker, at some point, call is transformed into a cast, which means there's no monitor on the worker process, and if it dies as the result, the call times out instead of failing otherwise.

[elbrujohalcon]

That is true, but we never found a good way around it. Do you have any suggestion on how to implement a call that's only dispatched to a worker (through wpool_queue_manager) when there is an available worker but still act as a gen_server:call from the outside without blocking the queue manager?
PRs are very much welcomed here :)

[ElectronicRU]

We could, before doing the cast_call, setup a monitor and then handle it in handle_info(). We can demonitor on worker_ready, too.
Edit: we would need to send a special message to the queue manager's caller and handle it there to raise a proper exception, of course, but such is the life.