See #132146-28 and #132146-32. Due to how we synced over content from one VM to another it might happen that there are results directory present that the openQA database considers non-existant hence causing an inconsistency and results that are never cleaned up. I can imagine that this might also happen in other cases so we should come up with an approach to ensure consistency.
- AC1: No job results are present in the file system for jobs that are no longer in the database
- AC2: Same as AC1 but also taking the archive directory into account
- AC3: AC1+AC2 are recurringly ensured on openQA instances (at least on demand)
- Create a script to do this kind of cleanup and run it once (before running it for real, make sure it would do the right thing by doing a dry-run first)