Parallel tetrahedral remeshing using TBB

Iteratively reshaping the cells of a 3D mesh to improve quality for simulations like finite element analysis — is computationally expensive: large meshes can take hours on a single CPU core. CGAL provides a widely used implementation, and GSoC 2025 built the infrastructure to run it in parallel, achieving a ~28% speedup on large meshes. Two problems remained: threads spent most of their time waiting to access the mesh rather than doing useful work, and the parallel version produced lower mesh quality than the sequential one. The 2025 code also never made it into the official CGAL repository. This project fixes both problems and introduces a new scheduling strategy: instead of rigidly applying all splits, then all collapses, then all flips, then all smooths, a unified priority queue ranks every candidate operation by its estimated quality improvement and always applies the most beneficial one first. This has the potential to reach the same final quality in fewer operations and has not previously been applied to tetrahedral remeshing.