Try to add simd support

this would maybe make nyasynth a lot faster at the cost of requiring a bunch of code in `sound_gen.rs` to be rewritten. also i dont actually know how to do this w/respect to getting the parameter modulation stuff to work with it. maybe consider doing per-voice simd instead??