Proposal: Add Mega L2F and L3F Scrambler

[ah1112]

Hi,

I was wondering if we could also have a last 3 faces scrambler (L3F) and last 2 faces scrambler (L2F) for Megaminx as those would be very useful as well.

For L3F it is very simple since we just need random RUF moves. I have implemented it here the same way you did for the S2L scrambler, and it should be ready to go: https://github.com/ah1112/cstimer/tree/megal3f

For L2F it is a bit trickier. We want a random state on the last 2 faces, which cannot be achieved with a 2 gen scramble. Really, we just want the same logic as src/js/scramble/mgmlsll.js but to not restrict the F moves. For last slot + last layer we can only have F' U* F since we don't want to break any other slots in the F face. But in L2F arbitrary F moves are ok, while still needing the restriction on R moves. I drafted some rough changes here, though it does not work very well, and the logic in the code is messy: https://github.com/ah1112/cstimer/tree/megal2f
I could make a separate file with the L2F logic, but you probably don't want unnecessary extra files.

Please let me know what you think.

[cs0x7f]

Currently, cstimer has the ability to solve megaminx states, meaning it can generate random-state scrambles for any desired state. However, the solution length can be very long, and I haven't found a good way to solve <R, U, F>. Your commit is not a random-state scrambler, it was just simple string replacement. Writing a solver isn't that easy.

[ah1112]

Understood. And you are referring to the L2F commit in this case. Yeah, I knew it was not a good commit. So what could be done for L2F? A separate file with the logic for the scrambling? I could write that if you like. Or do you have something else in the codebase that would handle this better?

And what do you think about the L3F commit?