I think this is min cycles, but not optimized for cost. I had a neat-ish 22 cycle solution, saved one cycle by rotating the output, but the other needed me to reroute the quicksilver, leading to this mess. Also, using the same molecule for both outputs just feels wrong.