So this was an interesting day and even if I'm committing this now, I
did the changes the day it was solved, but I didn't have time to write a
longer commit message. But I remember this day quite well, so I'm
writing them now.

So it turns out I was extremely lucky today. I lucked out on both parts,
in two different ways. The first mistake I made, which I made in both
parts, is that I started the "going right" starting point, with 1 step,
which is not correct. I did however start the "going down" with 0 steps,
which is the correct thing to do. And luckily both part one and part
two's solutions start with going down.

The second mistake I made was that I started the left and right steps
with 0 steps, instead of 1. This just had the effect of making it so I
had to do less than or equal to 2 instead of 3, but I wondered why it
was like this during the competition.

The last piece of luck I had was that I didn't read the final line, that
you had to move at least 3 squares, also on the last move! In the end it
didn't matter for my solution, but it easily could have!

The way I figured out most of these mistakes, was that I actually
thought I had found a bug in the puzzle. I got a path that was cheaper
than the answer on part two, and wondered how this could be. This was
after I fixed the "1 step on going right" bug I describe above. After
visualizing the mistake and other things, I made sure that the route
described was correct and went to reddit to make a comment. That's when
I read in the mega thread about the fact that you had to move at least 3
squares also on the last move and also the fact that other inputs had
been tricky with the starting with 0 steps policy.

Overall I was very lucky on this day, I think it could have taken a long
time to debug why my answer was not the answer, especially as it was
lower than the actual answer.

This was just cleaning up a bit, but it seems that we could improve this
more, by reading the reddit thread. I will try to do this, but right now
this is just a cleanup commit.

And with this, I have finally solved all the tasks in 2023. I was
reluctant to enter into this one, because I thought it would get rather
ugly, but once again, the input is very kind.

I was able to solve it, by noticing that there is no stones in the
starting row or the starting column. There are also no stones on the
edge of the map.

Using this, I partition the grid into 8 zones and I calculate how many
plots we get to for each zone. This is fast and runs in O(B) where B is
the size of the initial board, as we have to do a BFS search on the
board to figure out which spots are reachable and the cost of reaching
each space.

reading the reddit thread, it appears that it's possible to solve this
in some other way, using a diamond shape. I slept really badly tonight
so I'll look into that a bit later.

This finishes this part of the saga, I've learned a lot and I'm very
happy that I took this deep dive into this task!

There are cleaner ways to implement this, using complex numbers, but for
now I am ok with my solution.

This solution was already good, I just had to remove a few comments. I'm
proud of part two here.

This was just another round of removing unneeded imports and removing
comments.

This turned out to be very easy once you realized that we are not trying
to solve the generalized version of the problem at all, but rather
something specific to our input. It can be reduced to just a few types
of nodes, either too big, empty or normal nodes. After that it's just a
simple calculation of finding the shortest path to the tile before the
target data and calculating how many steps it would take to move this
data all the way to the left.

This is of course assuming, that we have no blockers in the top column.