Optimize M2 update loop: skip static doodads, incremental spatial index

- Split M2 instances into fast-path index lists (animated, particle-only,
  particle-all, smoke) to avoid iterating all 46K instances per frame
- Cache model flags (hasAnimation, disableAnimation, isSmoke, etc.) on
  M2Instance struct to eliminate per-frame hash lookups
- Replace full rebuildSpatialIndex on position/transform updates with
  incremental grid cell remove+add, preventing 8.5ms/frame rebuild cost
- Advance animTime for all instances (texture UV animation) but only
  compute bones and particles for the ~3K that need it

M2_UPDATE: 10.7ms → 2.0ms, FPS: 35 → 55-59

include/rendering/m2_renderer.hpp

@@ -174,6 +174,13 @@ struct M2Instance {
     std::vector<float> emitterAccumulators;  // fractional particle counter per emitter
     std::vector<M2Particle> particles;
 
+    // Cached model flags (set at creation to avoid per-frame hash lookups)
+    bool cachedHasAnimation = false;
+    bool cachedDisableAnimation = false;
+    bool cachedIsSmoke = false;
+    bool cachedHasParticleEmitters = false;
+    float cachedBoundRadius = 0.0f;
+
     // Frame-skip optimization (update distant animations less frequently)
     uint8_t frameSkipCounter = 0;
 
@@ -451,8 +458,14 @@ private:
     std::vector<std::future<void>> animFutures_; // Reused each frame
     bool spatialIndexDirty_ = false;
 
+    // Fast-path instance index lists (rebuilt in rebuildSpatialIndex / on create)
+    std::vector<size_t> animatedInstanceIndices_;   // hasAnimation && !disableAnimation
+    std::vector<size_t> particleOnlyInstanceIndices_; // !hasAnimation && hasParticleEmitters
+    std::vector<size_t> particleInstanceIndices_;    // ALL instances with particle emitters
+
     // Smoke particle system
     std::vector<SmokeParticle> smokeParticles;
+    std::vector<size_t> smokeInstanceIndices_;  // Indices into instances[] for smoke emitters
     static constexpr int MAX_SMOKE_PARTICLES = 1000;
     float smokeEmitAccum = 0.0f;
     std::mt19937 smokeRng{42};