nwharbert8-ui/brain-connectivity-wj: v1.0.0: NeuroImage submission

Adds the peer-review revision analyses for NeuroImage: Reports manuscript YNIRP-D-26-00107. Included in revision/: Length-matched within-condition floor (revision_floor_matched.py) Four sign treatments and direct stratified sign-flip rates (revision_sign_and_metrics.py) Net-versus-gross uniformity index and alternative similarity metrics Group-averaging artifact with subject-identity permutation control (revision_group_artifact_fisherz.py) Post loss-of-responsiveness trimmed analyses producing every value in the revised manuscript (revision_final_trimmed.py) Revision figure generation at 400 dpi (make_revision_figures.py) Substantive correction from the original submission: subject-level weighted Jaccard was previously tested against a perfect-match value of 1.0, which is uninformative because repeated measurement of a single condition does not return 1.0. The within-condition floor is 0.544. Against that reference the state-attributable share of architectural divergence is approximately 11 to 14 percent, not the approximately 50 percent the original framing implied. The effect remains robust (paired Cohen's d 2.2 to 2.8, p = 2.4e-07); its magnitude is revised downward. Python 3.10, random seed 42 throughout.