//! SIMD acceleration layer with runtime feature detection
//!
//! Provides optimized vector and matrix operations using platform-specific SIMD
//! instructions (AVX2/AVX-512 on x86_64, NEON/SVE on aarch64) with automatic
//! scalar fallback for unsupported architectures.
pub use SimdCapabilities;