Complete Run Artifacts for 'Certified Discovery on Real Grams' and 'Dark-Mass Metrology': Scripts, Execution Logs, Graded Scorecards, and Data Tables

Complete primary run artifacts for the two empirical papers of the certified-discovery trilogy: "Certified Discovery on Real Grams: an Instrument, its Limits, and its Laws" (DOI: 10.5281/zenodo.21819895) and "Dark-Mass Metrology: Pricing What a Certified Model Does Not Know" (DOI: 10.5281/zenodo.21819925). These are the execution records the papers' numbers were read from, not post-hoc reproductions: every experiment ran under locks frozen in the script docstrings before first execution, and every scorecard is shipped as graded, with misses owned in place. The record contains two archives: paper2_artifacts.zip (32 files) carries the fifteen Paper II scripts, including the full G-series instrument pipeline (g1_qualify through g5_lawI), their execution logs with construction-iteration disclosures and graded scorecards, the five data tables (adult_twotier.csv, g1_construction.csv, g2_results.csv, g5_results.csv, german.csv), the fixture, h1, and oracle studies with their result logs and the family-certification negative result, the figure script make_figs_p2.py with the three paper figures, and a merge manifest. paper3_artifacts.zip (43 files) carries the seven named Paper III programs (b2_public_run, collapse_verify, p3_family_probe, reannotate, sl6b_resolve, yield_ledger_sim, and the probe log) together with the complete run artifacts of the dark-mass measurement campaign: the illum_g6, vault_probe_g7, zero_dark_g8, and suppressor_atlas_g9 program sets with logs, scorecards, prescriptions, verification tables, and interactive figures, the b2 and family-scan records, and the same merge manifest. Each archive includes MERGE_MANIFEST_21820010.md documenting per-file provenance and a byte-level audit. All artifacts originate from the recorded research sessions; the full pipeline was additionally regeneration-gated on a pinned environment (Python 3.12.3, numpy 2.4.4, scipy 1.17.1, statsmodels 0.14.6, pandas 3.0.2, scikit-learn 1.8.0, xgboost 3.1.3), reproducing every graded scorecard value exactly. Public datasets (UCI Adult, Statlog German credit, wine-quality-white) are fetched from public mirrors with integrity assertions in the scripts; no restricted data is included. Archive checksums: paper2_artifacts.zip md5 6e2b9cb80c3aed630d8ce9fd1cf66309 paper3_artifacts.zip md5 0101ad64fee67541110465cce455bd37. The accompanying software package is archived at DOI: 10.5281/zenodo.21819947