to save on compute and time we could only scan "new" image hashes running on the cluster, instead of scanning everything.
to save on compute and time we could only scan "new" image hashes running on the cluster, instead of scanning everything.