Yet another better duplication setup allows for 2 less cycles of setup before beginning the main loop of the solve, as well as requiring a few less components to decrease cost. 
I have never before found this much optimization in potential in a single design. Without any drastic changes to the overall flow of the puzzle, only optimizing one section at a time, I've dropped 23 cycles. A part of that was assuredly due to me not seeing perfectly optimal ways to do everything at first glance, but even then this has been a remarkable improvement.