feat: simd128 support for wasm32

[mertalev]

Description

This PR adds simd128 support for jxl-rs when targeting wasm32. It does not include lcms2 integration for Wasm since this requires build toolchain changes and increases the scope of the PR.

In a local benchmark on an M4 Max, performance increased by 2.49x (8.9 to 22.1 MP/s), 2.02x (11.1 to 22.5 MP/s) and 1.35x (34.5 to 46.6 MP/s), respectively, for the progressive.jxl, bike.jxl and cafe.jxl test images.

[github-actions]

Benchmark @ da909d2

MULTI-FILE BENCHMARK RESULTS (8 files)
  CPU architecture: x86_64
  WARNING: System appears noisy: high system load (2.18). Results may be unreliable.
Statistics:
  Confidence:               99.0%
  Max relative error:        3.0%

Comparing: 7cf3a662 (Base) vs 86f9c2dd (PR)

File	Base (MP/s)	PR (MP/s)	Δ%
bicycles.jxl	7.984	7.971	-0.15% ±0.7%
bike.jxl	24.613	24.309	-1.23% ±2.4%
delta_palette.jxl	6.683	6.682	-0.01% ±0.4%
green_queen_modular_e3.jxl	8.768	8.728	-0.45% ±0.6%
green_queen_vardct_e3.jxl	23.453	23.389	-0.27% ±3.0%
lz77_flower.jxl	3.655	3.660	+0.14% ±1.1%
patches_lossless.jxl	2.880	2.872	-0.30% ±1.9%
sunset_logo.jxl	3.093	3.099	+0.19% ±1.1%