regularly scrub all the data stores of swh
Make sure we have background jobs that regularly/constantly check data integrity in all the SWH data sources:
- check hashes stored in the main postgresql storage (and replicas?)
- check objects stored in kafka
- check blob hashes for objects stored in all the objstorages (saam, azure, s3)
For example, doing mirroring tests, I found several blob objects in S3 that look to be corrupted (but original copies in the main objstorage are fine).
Migrated from T3841 (view on Phabricator)