Clean runs of our FR5 puck transfer work: pick, place, advance pocket. A stop between those actions can leave the saved pocket out of step with the physical cell. We have 128 planned. How should I design recovery around what is known, rather than trust that one number?