[andrescera]

• Add opt-in H1 v2 RSS evaluation after a ten-minute warm-up and thirty-minute measurement, preserving frozen v1.
• Derive sixty-second capture samples and retain the selected rule's version and hash.
• Add thirteen regression tests for RSS windows, preserved thresholds, capture duration and provenance.
• Document selecting either rule and retaining historical v1 verdicts.

Verified locally: native build, 112 CTest cases, 101 Python compat tests, H1/G5 selftests, workflow contracts, receiver-only install, and clang-tidy with the repository configuration.

Live capacity requires new measurements and independent review.