Release v0.7.6

Features

• Unified Math Operators: Introduced standardized maths::clamp, maths::saturate, and scalar/vector maths::lerp overloads in maths/ops.h and maths/f3x1.h to streamline interpolation and bounding across CUDA kernels.
• Robust Normalization: Added maths::normalize_safe with configurable fallback directions and numerical thresholds to eliminate degenerate division issues across mesh ray traversal and feature calculation.

Refactors & Build

• Call-Site Consolidation:
  • Replaced ad-hoc clamping, lerping, and saturation logic across Dual Contouring (dc.cu), Dual Marching Cubes (dmc.cu), Marching Cubes (mc.cu), Asymptotic Marching Cubes (mca.cu), Marching Tetrahedra (mt.cu), Marching Tetrahedra Grid (mtg.cu), Mesh BVH (mesh_bvh.cu), Triangle Mesh (triangle_mesh.cu), and Triangle primitives (triangle.h).
  • Cleaned up unused residual functions in qef.h.
• Package Version: Bumped version to v0.7.6.

Examples

• Streamlined CUDA kernels ensuring consistent numerical behavior across all isosurface extraction and spatial hierarchy traversal routines.