First off I want to whine about the input being chiral, and the wrong kind of chiral. I think everyone that's solved this puzzle knows what I'm talking about, you need to spend a bunch of effort fixing up the input one way or another.

I have made so many solves and so many attempts. My 7th solve was the best I had for a LONG while at 367. After many many many fruitless attempts and approaches I managed to find a solve that beat it... by 1! One. It was enough of a different solve that I was able to play around with it for a while and get things down to 361, though.

I do wonder sometimes, if the real massive improvements aren't hiding somewhere in my -other- solutions, the ones that seem like they are obviously worse. Every time I make a new solve and compare it against my current best, I -am- comparing a solve I tried to optimize as hard as possible with a brand new one I just made. But more likely I just missed better approaches entirely.

Shoutouts to arm #2 which is 1-length and very deliberately placed to not eat up as much area as it used to, and I also played around with the output sequence of arm #5 to reduce the area. Other than that it's a pretty simple solve I think, constructing the product by halves.