trackless cost is the metric i didn't know i needed. this was fun to optimize. i had a 120g solve with two piston arms that i thought was good, but then i realized i could use only one piston arm and save 20g. i can't prove that cheaper is impossible, but i certainly won't find it. then on the last day, i got a wave of motivation and managed to save over 100 cycles. maybe i could save more, but i've submitted an unhealthy number of times already, so i'll leave it here.