My instructions journey began on Tuesday afternoon.  After reaching a reasonable stopping point on cycles, where my lowest instructions was something in the 200s, I made an 83 instruction solution with a completely different method.  Assembly happened in the top left chamber.  I messaged zorflax that it felt like I had completely forgotten how to do instructions.  Then all of the following numbers were my best at some point along the way:
81 78 73 68 67 65 64 51(!) 49 47

47 is a lot lower than 83.  How did I get there?  I explored lots of different workloads per chamber, but what ended up working the best was the exact same as I used for cycles.  That is, inputs in one small chamber, animismus and chunky mors in the other, and assembly in the large chamber.  For a while 1/3P and 1/6P were competitive with each other, but when I was able to get the conduit arrivals synced up in an ideal way, 1/6P became a whole lot better (this was the plunge from 64 to 51) and left 1/3P in the dust.  I did tinker with 1/12P and 1/18P but it didn't appear to save me any instructions, and would obviously be slower.

Getting this layout to its fastest possible machine period is nontrivial.  The silver wants to arrive too early.  It needs to stall by more than an entire tape loop before dropping through the conduit, to be after the vitae has had a chance to swing and push.  At first I cursed the need for a handoff between arms 7 and 8, but this turned out to be essential to stall by the necessary amount.  So now we have unusual tapes on those two arms consisting of heavily delayed drops.

None of the arms themselves require period 10 (I have a version where in the absence of the period override, the machine period would be 8).  But there is a very complicated dependency chain due to this stall, that fails if the period is lower than 10.  So that makes this a 1/6P10 solution.  To my knowledge, there is no secondary or tertiary left on the table.