Retry or failed for all in deduplicated jobs that fail. #4533
Description
We could maybe introduce a flag for this, allowing the user to set the desired action. I think both routes could be reasonable. Fail for all, or retry as long as there are similar jobs.
Maybe, it could even be related to the type of error or boefjes what the desired course of action should be. timeouts are probably more transient than parse errors on our input object for example.
Originally posted by @underdarknl in #4482 (comment)