Changes in version 0.1.0 (2026-10-08)                  

  - Initial release.
  - Expected-gain model for repeaters fitted by marginal ML, robust to
    preknowledge (rt_fit()).
  - Posterior-predictive evidence: score gain, exposed-item performance,
    response speed (rt_evidence()).
  - Risk index calibrated by parametric bootstrap with explicit
    false-positive rates (rt_risk()).
  - Raw-gain baseline (rt_raw_gain()) and known-truth simulation
    (rt_simulate()).