I had long notes that got eaten by the site, so have point form notes instead (feel free to interject to explain what I'm saying to the stream):

- Once a salt, always a salt. At least in min cost.
- I thought there would be shenanigans to minimize fire grabs, but given how spectacularly useless salt is at min cost, my two best ideas for shenanigans are worthless.
    - First was to convert fire to salt just to do a few grabs, then convert it back.
    - Second was to use salt as a wand. Not technically ruled out, but I'll be surprised if it's a massive factor.
- I really like how this puzzle works out at min area.
- The layout is quite constrained, I'll be surprised if there's a better one.
    - Unbonder needs to be near input or else you lose suppression on the first input.
    - Triplex next to input would require a lot more shuffling for suppression.
    - Both unbonder and triplex need to be double-accessible by the arm, barring absolutely unhinged wand shenanigans.
    - I didn't experiment doing it in the other direction, which means the other direction is probably better.
- There is one and only one secondary optimisation: when handling the fifth input, the waste chain is rearranged to let a move be done by salt rather than fire. -1 fire grab, +2 salt grabs.
    - This is needed because the unbonder is in the way of bonding the output.

I look forward to everything I said being proven wrong.
- The shuffling at the start with salt is not an optimisation.
    - This is just a setup that makes it work 1T with both a waste chain (prone to collision) and none (doesn't suppress input without extra work).
    - The cost of 1T is only 2 salt grabs, and 2 tertiary points is worth not having to write a 6T tape.