Implement a debug flag to give more insight in crashing workers

See https://github.com/josdejong/workerpool/pull/197#issuecomment-715530568:

> I think it may be helpful to add something that notifies the process has been forcibly shut down. a consumer may or may not care, but in my case, it could cause flaky coverage reporting... and/or a configurable timeout

And we can probably think of more information that can be logged when `debug` mode is on.