Mounting plates counted twice after reviewed retries

Each plate has an identity. Restart changed the attempt number, not the plate label.

Then show attempts separately and derive accepted plates from the authorised disposition for each plate. A pass row on its own may not be the current decision if another check was reviewed afterwards.

Hazel, worth spelling out what happens when a reviewed fail follows a pass. A simple 'has ever passed' filter would make the number prettier and still wrong.

Quality uses the latest authorised disposition, not any pass. We found 224 accepted plate identities in the reviewed export.

Reconcile the rest of the physical batch as well. Accepted, awaiting a decision and rejected should account for the plates, while the attempt total can legitimately be larger.

224 accepted, nine awaiting review, four rejected: all 237 plates accounted for. Thanks, Hazel. Attempts are separate now.

Have they tested the rebuilt count through restart and a changed authorised disposition? Reproducing today's export alone won't expose the old counter incrementing again.

Yes. Restart replay stays at 224; changing one accepted test plate to reviewed fail reduces the test total to 223. Its attempts remain visible. Installed dashboard reconciled to the unchanged batch at 224.