Add macro-controlled SIMD support (SSE/NEON) to math library

[Copilot]

Adds compile-time SIMD acceleration for Vector4 and Matrix4 operations, with platform detection for x86 (SSE) and ARM (NEON) and scalar fallback.

Changes

• cmake/options.cmake — SKY_MATH_SIMD option (ON by default)
• cmake/configuration.cmake — propagates SKY_MATH_SIMD as compile definition
• MathSimd.h — platform detection macros: SKY_SIMD_SSE, SKY_SIMD_NEON, SKY_SIMD_ENABLED
• Vector4.h — __m128/float32x4_t as union member alongside float v[4]/{x,y,z,w}
• Vector4.inl — SIMD paths for +=, -=, *=, /=, Dot, Normalize (with Newton-Raphson refined rsqrt)
• Matrix4.inl — SIMD mat×mat and mat×vec multiply
• test/core/MathTest.cpp — tests for Vector4 arithmetic, dot, negate, and Matrix4 multiply

Design

SIMD type lives in the Vector4 union for zero-cost access. Operations use direct intrinsics with three-way #if SSE / #elif NEON / #else branching:

inline Vector4& Vector4::operator+=(const Vector4& rhs) {
#if SKY_SIMD_SSE
    simd = _mm_add_ps(simd, rhs.simd);
#elif SKY_SIMD_NEON
    simd = vaddq_f32(simd, rhs.simd);
#else
    x += rhs.x; y += rhs.y; z += rhs.z; w += rhs.w;
#endif
    return *this;
}

Normalize uses _mm_rsqrt_ps / vrsqrteq_f32 with one Newton-Raphson iteration for precision.

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.