Struggled on the fact that "rounding *down* to the **nearest** integer" doesn't mean `round()` it means `floor()` for part1.
Got the idea in the back of my mind this is the thing that could be heavily paralellized.

For part2 I found a solution that would take 12h in brute force.
Removing a list creation I got it down to 3 hours (`two_bf.py`). I let this run in the background.
Using numpy shenanigans I got it down to 20 minutes.
It worked before I could do further optimisation.

I have no fucking idea what could be the optimisations here,
and not super interested in figuring out. I'll spoil myself.

---

OH

🤚: buyer × buyer's sequences
👉: all possible sequences × buyers

Yeah alright