v1.0.0 — first public release

Five-stage screening pipeline over the Materials Project: pull + hard filters →
economics/synthesis scoring → empirical gap correction + low-PBE re-pull →
ML gap model (matbench_expt_gap, CV MAE ~0.4 eV) → family novelty rules,
hard vetoes, tiered shortlist.

Headline: Ba3Nb2Se9 — ML-predicted 1.35 eV vs 1.30 eV measured
(Inorg. Chem. 2013, doi:10.1021/ic4013763); zero PV-absorber literature.

See SHORTLIST.md for the tiered candidates and WRITEUP.md for methods,
validation, and failure cases.