REG-059 — after (buyer-facing release-state blocks collapsed; the rule is now a gate)
=====================================================================================

Row:        REG-059  [P3-2]  DOC  P3  "Buyer-guide duplication"
Transition: SEED -> FIXED

(1) What was duplicated, counted:

    docs/BUYER_GUIDE_US.md         3 restatement lines collapsed -> 1 line
    docs/PRODUCT_BRIEF_US.md       3 restatement lines collapsed -> 1 line
    docs/COMMERCIAL_STRATEGY_US.md 2 restatement lines collapsed -> 1 line
    docs/PROSPECTUS.md             2 restatement lines collapsed -> 1 line
    docs/FAQ_PROCUREMENT.md        3 restatement lines collapsed -> 1 line
    ------------------------------------------------------------------
    13 restatement lines removed, 5 pointer lines with the boundary kept.

    The surviving line keeps what matters and drops the copy:

    **Release baseline:** checked-out source baseline `v5.0.0` with fourteen
    synchronized anchors, published 2026-09-16 on every surface except PyPI
    `aegis-latent-core` — per-surface readbacks, the latest published release and
    the historical baselines are stated once, and only, in
    [`docs/RELEASE_STATUS.md`](RELEASE_STATUS.md) §1.0–§1.1, which is the only
    document that establishes publication state

(2) The defect the copies were hiding. With the pointer rule in place, the new
    check failed on the real corpus immediately:

    missing-status-pointer: docs/PROSPECTUS.md:1 — no pointer to docs/RELEASE_STATUS.md

    Inspecting it: the baseline note told readers to "verify the `v4.1.1` tag,
    GitHub Release, PyPI and npm artifacts, OCI digest, signature, and
    attestation" — two releases behind the recorded external baseline (v4.1.2)
    and pointing at nothing. This is the failure mode REG-059 predicted, found in
    the corpus rather than argued about. Its note now states the boundary and the
    pointer (the capability list is untouched).

(3) The gate, and the proof it fails closed:

    check_buyer_document_baselines() in scripts/verify_docs.py flags
    - duplicate-baseline-restatement  (`**Source|External|Historical external baseline:**`)
    - restated-readback-digest        (a v4.x.y line carrying a 40-hex SHA)
    - missing-status-pointer          (no RELEASE_STATUS.md reference)

$ # planted a duplicate baseline line in docs/BUYER_GUIDE_US.md
$ python scripts/verify_docs.py --root .
    duplicate-baseline-restatement (1):   -> exit 1
$ # reverted
$ python scripts/verify_docs.py --root .
verify_docs: PASS (0 findings)   -> exit 0

(4) Regression tests (both directions — silent on the pointer form, failing on a
    restatement):

$ pytest tests/test_documentation_verifiers.py -q
31 passed in 0.91s

(5) All four documentation gates after the edit:
verify_docs: PASS (0 findings)
verify_claims: PASS (102 claims, 0 findings)
verify_links: PASS (1294 relative links and anchors resolved)
verify_documentation --strict: status=PASS errors=0 warnings=0 required_files=27
