[raymath] Added SSE to MatrixMultiply()

[mcdubhghlas]

I saw your LLM challenge on Twitter and decided to try it. I had to edit it a bit to make it work, but hey. It runs and it does what you need it to. I saw #5316 and was like "oh yeah."

I am getting roughly half a second per 200000000 ops, with a range of 2.5ns - 3.2 ns per op. The after picture is roughly one second per 200000000 ops, with a range of 4.6ns - 5.1ns per ops.

I used an LLM to generate the test: https://gist.github.com/mcdubhghlas/53f780c35482aa87713f34ded941a67d

It was compiled using this: cc benchmark.c src/libraylib.a -Isrc -O3 -march=native -lm -lpthread -ldl -lrt -lX11 -o bench

Let me know if that is insufficient, I genuinely was too busy to read over documentation to see if you had a test suite -- That said, I love you what you do Ray. Keep up the good work man :)

[raysan5]

@mcdubhghlas did you try this change with some raylib examples, to verify that nothing is broken? Specially with 3d examples...

[raysan5]

@mcdubhghlas thanks for the improvement, I'm reviewing it to allow users select the SIMD option is desired