Synthetic data for forward-conditioned means and null calibration in onset-selected ringdown tests

Saved configurations, calibration and test scores and profiles, paired processed synthetic arrays, source-data dictionaries and numerical replay supporting Forward-conditioned means and null calibration in onset-selected ringdown tests. There are 199 calibration pairs, 200 initial test pairs and 200 continuation pairs. Each continuation pair is reused across six cases and both analysis arms, not 1200 independent noise pairs. These are processed white Gaussian simulations with frozen physical inputs, not measured gravitational-wave detector data or event-level significance estimates. The 599 distinct saved noise-pair hashes establish byte-level distinction, not an empirical proof of stochastic independence. The current archive preserves all 34 original scientific files and adds explicit license notices, archival documentation and reproducibility guidance. First-party synthetic data and generated figures are CC BY 4.0; packaged first-party replay code is MIT. Dependency and third-party rights retain their own terms. No manuscript or preprint is included. Numerical verification and publication of supporting material do not constitute scientific peer review or journal acceptance. Use ringdown-benchmark-data-v1.0.0.tar.gz, version 1.0.0; paired software has DOI 10.5281/zenodo.23598653.