Implemented support for 128 bit tmhm learnsets.

[StephenLynx]

First of all, the gTMHMLearnsets was changed to an array of 4 32 bit numbers.
Then the macro was changed to take in account tms that belong to the second half.
Consequently the function to check if a mon can learn a tm move was adapted.
I also added the defines necessary to both the list of tm items that take the name of the move as the list that associates tms with moves.

Stephen Lynx#9875

I remade the PR because the previous one was based off the wrong branch. I now pulled gen 8 learnsets and kept any learnset that wasn't present on pory. The issue being that most mons lost the ability to learn hms.

[StephenLynx]

I decided I'll pull movesets from gen 7 where possible and then from gen 8 where the mon is not found on gen 7. The reasoning is that the way I do, every mon that is not present on gen 8 won't have A LOT of moves, because it's limited to tms 1-50, then the ones that ARE on gen 8 won't have HMs because of the move purge.

[AsparagusEduardo]

Not a fan of defining custom TM spreads instead of using either Gen 3 or 4's. IMO, these choices should be up to the user.
As discussed on the discord, keeping the bit array approach becomes an issue the more TMs one has. If someone were to try to replicate SwSh's 200 TM/TR list, they would have to deal with this once again.
Splitting the macro to have 2 arguments makes things way too messy both format-wise and usability-wise.

The one thing I like is making CanMonLearnTMHM use CanSpeciesLearnTMHM, as it's redundant to have pretty much the same function copied over twice.