After the first 2 loops, nothing happens until around cycle 3800.

This solution works because with careful placement, both copies of pictric acid can be picked up on adjacent loops, so letting it iterate through all track locations will pick up the parts in the correct order. Using extra track for more free space and breaking the bond on the second pictric acid at the start of the loop reduced the number of collisions I needed to deal with, without that I don't know that I would have had the willpower to finish.

When I first thought of this I was very confident it would do well, but the more I've considered it the worse I think I did. None of the individual parts of the solution are hard to think of, since it's the same construction method as with a 4 grab solution, and while a 2 grab solution was much easier to implement it's not much of a jump to think that 1 grab might be possible. Everything past that is figuring out the implementation, but past weeks have shown that lots of top players are great at that, so now I think I'll only place well if some players don't think implementing 1 grab is worth the effort. I know I could reduce cost on this by removing track, but it took too long to get this to work for me to be willing to do it again but with more collisions.