Turns out fire-quicksilver hex is more workable than fire-copper or all-reagents to build into reasonably fast half assembly. The big problem under this motif is that projection happens close to the product quicksilver but the resulting silver must bond between fires which is far away from it. Separate silver and copper arms help this storage, and makes hex arm wait-free nicely.

The triplex-merge-halves part bleeds area, but there's only so many options when you have one arm and only 9 free cycles between second half's grab window and first half's grab window (in other words, processing first half faster doesn't help R at all).

I had hope for early-triplex build order, but turns out all my sub-400 is triplex-last, hexarm or not. Maybe handling the center is just too awkward and building a half is just too batchable.