Add checkpointing on storage_checker to avoid rechecking objects at the beginning of ranges again and again
Staging:
* [ ] apply `swh/scrubber/sql/upgrades/4.sql`
* [ ] upgrade package on workers and stop all workers
* [ ] start one worker
* [ ] wait for a couple of `Processing %s range %s to %s` lines
* [ ] restart it
* [ ] check it is not processing the same ranges
* [ ] start other workers
Production:
* [ ] apply `swh/scrubber/sql/upgrades/4.sql`
* [ ] upgrade package on workers
* [ ] restart all workers