Taking stock -
Learning fusion was fun, constraints are really powerful, but they work against you if your datum is wrong (or moving in the hole as you measure 😖)
Being able to quickly and consistently print very thin test pieces is a game changer.
Approach that eventually worked was to disassemble the whole thing, rather than measuring from the bolt-heads - actually bolt the token against the part through the known bolt locations (once they were right.) Start with way bigger holes than needed (especially around the posts) and tighten down incrementally as everything else locates properly.
Also - minimal constraints first - add them to position elements as needed.
And don’t treat a wonky head location as a datum point.
