[Task]: Re-run previously failed background jobs that never had finished/success set

[tw4l]

Description

As a consequence of #3044, a number of background jobs that failed never had their finished or success fields set. This means they cannot currently be retried.

We need a programmatic way of identifying all such jobs, setted a finished datetime and success: False in the database for them, and then use the existing API endpoints to retry all failed jobs. This could be handled via a database migration or scripts, tbd.

Context

No response

[tw4l]

Given that some of the failed jobs shouldn't have been created in the first place (e.g. replication for failed crawl WACZs) or will no longer be relevant (e.g. replication for since-deleted crawls), I think the best approach would actually be to handle this in two steps:

1. A migration to find older background jobs that failed but never had finished or success set and set those fields
2. A migration or script to look for any crawl/upload/profile files that should be replicated but are not yet, and start new replication jobs for those