Replication and data package: Engineering optimal open addressing without reordering for high-load hash tables

This is the replication and data package for the paper "Engineering optimal open addressing without reordering for high-load hash tables." It contains everything needed to reproduce the figures and tables in the paper from raw measurements, and to re-run the benchmarks from source. The package includes: the five publication figures; aggregated and per-seed measurement data from both the primary Apple M4 environment and a secondary container environment (throughput, latency percentiles, probe counts, churn drift, critical-amplitude grids, and overflow-occupancy trajectories); the C++ benchmark driver and the diagnostic programs for the amplitude sweep, grid bisection, overflow census, and deletion stress tests; the data-collection and figure-generation scripts; and a build manifest documenting the exact environment and reproduction commands. Probe-count results are cross-platform deterministic and verified bit-identical across instruction set, SIMD backend, compiler, and operating system. Three external libraries (ankerl::unordered_dense, robin_hood, and IcebergHT) are bundled as comparison baselines and retain their own licenses; see THIRD_PARTY_NOTICES.md. Original code is released under the MIT License and the data and figures under CC BY 4.0; see README.md for full details and a quick-start guide.