Only appears possible from the current view. The designer is checking the actual mounting under the approved setup procedure. I've kept that as a question, not a confirmed fault.

The check confirms it can mount reversed. The designer also found that the other family's insert can share the mounting pattern. We're comparing physical discrimination options and a setup-to-recipe confirmation scheme with controls.

Carry both findings into the candidate review. Preventing reversal may do nothing about the wrong family, and correctly identifying an insert does not establish that it's seated in the intended orientation.

What happens if changeover stops halfway and the previous confirmation remains stored?

The proposed scheme invalidates the previous setup confirmation when changeover begins and keeps an incomplete changeover held. That is design intent at present, not a tested installed behaviour.

Have the normal changeover operator review it, including loss of an identity reading and a correct recipe with a physically wrong insert. The test should show how uncertainty is handled, not only how the happy path reaches confirmed.

Those cases are in the joint review, with the normal operator included. The mounting changes have not been made yet. We have confirmed the two mix-up possibilities and defined candidate checks, but automated use remains unauthorised.