lctriage — experiment outputs for "Persistence, Not Consensus" (TESS sectors 1, 2, 13)

Experiment outputs behind every table and figure of Persistence, Not Consensus: Measured Triage of Unsupervised Anomaly Rankings of TESS Light Curves, produced with lctriage 1.1.0 on TESS sectors 1, 2 and 13. Kept apart from the catalogue so that scores computed on synthetically modified light curves cannot be mistaken for measurements of real targets. Contents: injection truth tables (sectors 1 and 2: type, parameters, seed, amplitude stratum of every injected signal and artefact); per-target ranks of every detector, windowed ranks and per-window feature matrices of the clean and injected runs; contamination diagnostics harvested from the raw products (CROWDSAP, FLFRCSAP, flux–photocentre and flux–background correlations); the paper's tables as CSV with counts and bootstrap intervals; the test–retest results on the targets common to sectors 1–2 and 1–13; the preprocessing baselines; the row-order and tie-policy audit; the dossiers and pixel-level checks of the inspected candidates; and the analysis plan frozen before any light curve of another sector was read, with its dated deviations and its SHA-256 digest as printed in every run log. The analysis plan states its own provenance: frozen before any light curve of sectors 2 or 13 was read, its SHA-256 digest printed at the start of every run log. The experiment outputs include JSON provenance and verification records where applicable; run settings (tie policy, excluded features, seeds, window edges) are recorded in the corresponding provenance metadata and run logs. CC BY 4.0.