docs(p037): record A2.2-S cumulative governed evidence

One governed measurement of the whole A2 treatment, executed via the
reviewed evidence orchestrator (e399575,
tooling branch claude/p037-a2.2), against population T
(4a8e658) and baseline R
(5fd6bfa), at treatment
dab3db1.

Result: FACTS MOVED · MOS UNCHANGED · VERDICTS UNCHANGED · accepted=true,
is_evidence=true, state=accepted, all eligibility predicates met.
138 documents (1 repo + 137 corpus), 47 changed (every change classified
guarded_facts/guarded_functions only, 0 unexpected), 91 unchanged. Zero
Python or Rust layer mismatches (lowered/summaries/verdicts) between
baseline and treatment; zero Python/Rust cross-engine disagreements;
zero verdict-snapshot cross-engine disagreements across 137 files.

Evidence-only: no instrument, treatment, driver, or unrelated doc
change. Files match exactly what the orchestrator published to
--out, copied verbatim.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011wKL9pEApQTK4dE3eXv8g8

Author: PhysShell