Implement Super Simplex noise

[mystise]

Implemented in 2D and 3D, no 4D implementation was available to port, so it was left out.

The maximum values were determined in Mathematica, and as such should be as exact as possible, so the output range should be [-1,1]. The Mathematica script used to produce the maximum values is included both in the comments in the new example, and also in the find_maximum_super_simplex.nb file.

The example produces the necessary lookup tables, and was originally going to produce the maximum value before the computation was moved to Mathematica, since the Newton's method code I wrote (and then deleted) was producing incorrect values which were provably not maximum.

Closes #161

[Razaekel]

Awesome! Thanks for adding this! That must have taken a while to figure out the conversion from the original java implementation, and dealing with the issues that came up.

My only request at this point would be that the find_maximum_super_simplex.nb file be removed. I know that I personally don't have a license for Mathematica, and without that, it would be hard to prevent the code from rotting over time. Ideally, the code would be implemented in Rust, which would make it easier to support, but I understand if you don't want to deal with translating that.

[mystise]

Works fine for me. If there were an open source alternative to Mathematica I'd also be happy to convert it at a later date. I tried to convert it to Rust, but for some reason it was returning provably wrong results. I can try again later, but for now I'll just remove the Mathematica file from the repo.

[mystise]

Request re-evaluation