For JGI, it's not impossible for someone to have multiple plates in a single submission. And in this case, the well IDs for samples within a submission may repeat. So we can't add validation that makes repeat values invalid (@pkalita-lbl , correct me if I'm wrong).
However, if there's only 1 plate, the well IDs shouldn't repeat.
Proposal is to have the transformation script check for repeat wells and provide a warning / flag when it happens. Then the reviewer can check if repeat wells are expected (multiple plates) or a mistake.