at some places (borg check --repair) we kill repository objects when we encounter IntegrityErrors on them.

for many cases this is the right thing to do - if an object is corrupt, it is not useful any more.

but when reading the LoggedIO code, i got the impression that some of the IEs raised there could be due to:

• corrupt repository index just pointing at some bad segment/offset where no object starts
• corrupt repository index pointing at a valid object, but for a different ID (maybe extremely unlikely, but who knows?)
• too large objects, https://github.com/borgbackup/borg/blob/1.1.0b4/src/borg/repository.py#L1175
• other cases?

https://www.bountysource.com/issues/44509028-review-integrityerrors-especially-when-they-kill