RRHM Open Lab — sealed prediction package for Staveland et al. 2026 (staveland-lock-v4)

Archived snapshot (git tag staveland-lock-v4) of github.com/morrocwi/rrhm-open-lab: the computational companion to the RRHM research program. Seals targets/staveland_2026 — an operationally frozen, falsifiable prediction about the public data of Staveland et al., Nature Communications 2026 (doi:10.1038/s41467-026-70287-5): a task-geometry escape-recoverability margin is predicted to add out-of-sample information about turnaround timing (recorded prior 0.45; binary failure criteria in the frozen text). Includes an independent reproduction of the paper's turnaround-by-reward coefficient (0.1114 vs original 0.1115) from the publisher Source Data, and a one-page prospective task modification. Code MIT; documents CC BY 4.0 (see LICENSE).