In some circumstances, import loop stops

[targos]

I've seen two cases:

• Import stops completely (no more logs)
• Import continues but doesn't find files anymore (when to_process is not empty)

[targos]

I think it's related to NAS storage restarting. We should try to reproduce that in a VM

[targos]

I think we should do the import itself in a child process. This will have the interesting side-effect to not require a restart of the import process when the config is updated.

[targos]

Other side-effect: if the import process crashes for some reason (for example running out of memory), we would be able to catch this and move the file to errored. Currently it tries to import it over and over