Add terrain footprint cutting benchmark for Mesh3D_Difference_Performance Records the conforming-Delaunay to ear-clipping change from DiGi.Geometry#2: ~1.9x on a lattice finer than the buildings, ~3.3x on the coarse lattice where remainders carry holes, and a dense terraced case that previously threw ConstraintEnforcementException. Includes a benchmarking note on why every figure is measured with the Fact run alone. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Refresh the Polyhedron closure benchmark for the weld-free predicate Replaces the section in place rather than appending a second run, per the benchmark guideline. Compares four implementations on identical inputs - the previous vertex welding, the two designs proposed on DiGi.Geometry issue 1, and what shipped - and adds the monotonicity sweep, which is the correctness result the timings exist to justify. Records the two hash traps that decide whether the fast path fires at all: the cell index has to be rounded rather than floored, because flooring puts the boundary on every exact multiple of the tolerance where clean geometry sits, and the ordinates have to be mixed one at a time, because exclusive-or is symmetric and a polygon on a circle carries the vertex with its ordinates swapped. Links Benchmark from Home, which the guideline requires and which was missing. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Add Polyhedron closedness benchmark results
Add Benchmark page with Average vs Centroid results Document the vertex-mean vs area-centroid scaling benchmark (DiGi.Geometry.Planar.Query.Average vs Centroid) with machine spec, fully-qualified method names, Release + Debug tables and analysis. Link the page from Home. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>