Multi-field advection support

Summary

Adds the ability to advect multiple scalar fields sharing the same velocity in a single WENO_step! call. Useful for problems involving multiple chemical components, tracers, or any set of fields transported by the same flow.

Usage

Each field gets its own u_min/u_max bounds for the Zhang-Shu limiter. Fields are advected sequentially, reusing the internal fl, fr, du, ut buffers — no additional memory is allocated.

Changes

New multi-field WENO_step! overloads accepting Tuple{Vararg{...}} of arrays:

• time_stepping.jl — CPU 1D
• time_stepping.jl — CPU 2D
• time_stepping.jl — CPU 3D
• KAExt.jl — KernelAbstractions 1D/2D/3D
• ChmyExt.jl — Chmy 1D/2D/3D

Tests:

• test_multi_field.jl — equivalence tests verifying multi-field WENO_step! produces identical results to calling single-field WENO_step! individually (1D/2D/3D CPU + 1D KA CPU)

Docs & examples:

• README.md — added multi-field section with usage example
• GettingStarted.md — added multi-field documentation
• 2D_multi_field.jl — example advecting 3 chemical components

Design decisions

• Sequential loop over fields rather than batched GPU kernels — simple, zero memory overhead, works on all backends. Batched kernels can be added later as an optimization if needed.
• Tuple{Vararg{...}} signature instead of NTuple{NF, T} to avoid Aqua.jl unbound type parameter warnings from keyword-args desugaring.
• Per-field u_min/u_max as Tuple{Vararg{Real}} — each field can have independent limiter bounds.

Tests

All 172 tests pass (including 9 new multi-field tests and all 11 Aqua checks).