Day 24: Use cycle finding for part 1
Day 25 further optimizations
Day25: Remove superflous commands
Day25: output more like other parts
Day 24: Use bitmap for part 1
Day 22 numbers fit in i128
Better instructions day 21
Cache distances to 'interesting' points in the maze. Search that
instead. Solve p1 and p2 in parallel.
Solve day 20
There could be some optimizations to do, but this goes in 200ms.