It took about 8 hours tinkering to make a 2.5 width machine that worked, then another 8 hours to make one that was actually cost-efficient via cutting out looping capacity and then a lot of wasted track (I hope someone else submitted a looping design, b/c the design compromises looping forces are interesting if quite expensive). This got me down to 200G with two four-track piston arms, two debonders, and two bonders. I thought that was min cost. It took a lot of lead juggling to ensure at least one arm could work at any given time, but it worked.

And then I realized "wait, can I might be able to remove a debonder." This concerned me, but reworking the arm programing for arm 1 wasn't *too* bad so I just did it and submitted that. And then, this morning, I had a thought: wait. If I can go down to 1 debonder, can I get only 1 bonder? You can't perfectly invert debonding to bonding to get the same result (due to accidental bonding being possible) but I did figure out a way to position a bonder so that this pivot-based product assembly came together nicely, so I had to implement it. If there's still cost savings to be found, it's in the possibility of some work order that allows you to save a track - this seems more possible in piston 2 than 1 due to the greater room to maneuver, but I'm not sure how.