Send an error event when Worker has failed after all retries

Some customers needs to handle exception when a given worker failed.
Work exception can very different based on the work type:

• like issues with filesystem
• timeout
  etc.
  They want for instance listen to an event to triage the document and update its status, or move the document and also send an email.
  The current AbstractWork#run simply raise an exception when all retries have failed.

We need a new default event workFailed fired once all retries have failed and attach the stack trace by default. We should also let the custom Work implementation override:

• what is being send in the event so the event can be enrich with more details of the root cause than a simple error massage