Act on the remaining PRISM#2 review items - #13

Description

Acts on the remaining open items from the PRISM v0.1.0 review (#2).

Branch protection is now provable, not asserted. The review's open item was
right: tests/platform/test_branch_protection_contract.py reads source files,
so it shows intent rather than the live GitHub setting.
configure_branch_protection.py gains --verify, which reads the live
protection and diffs it against the policy the script applies, exiting
non-zero on any mismatch:

python tools/ci/configure_branch_protection.py \
    --repo AI-ModCon/PRISM --branch main --verify --review-count 0

Run against both repos today, every field matches. Anyone with read access can
reproduce it, and it can go in CI — which a screenshot cannot.

Two text fixes from the review: trainer_grpo.py said "meaned"; the NCBI
reference is trimmed to https://eutils.ncbi.nlm.nih.gov, which resolves,
rather than the /entrez/eutils path, which 404s.

A claim in our own docs was wrong. docs/models/intern-s2-preview.md said
Qwen/Qwen3.5-35B-A3B-Instruct is "gated on Hugging Face" and that its
HTTP 401 is a sign-in prompt. It is not. A gated repository serves its model
card with HTTP 200 and gates only file downloads — a known-gated repository
returns 200 on the same anonymous request — so 401 means the repository does
not resolve. The rest of the family does (Qwen3.5-9B, -4B, -35B-A3B,
-397B-A17B), so the -Instruct suffix looks like the error and
Qwen/Qwen3.5-35B-A3B is the nearest checkpoint that exists. The note now says
that, with the evidence table, and is explicit that the exact model behind
those defaults is unconfirmed.

The files that cite it are deliberately untouched: vendored upstream code kept
byte-identical for re-sync, excluded from lint and formatting in
pyproject.toml, and CONTRIBUTING.md forbids annotating it.

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• Documentation update
• Infrastructure, CI, or tooling

Compatibility

• src.* import paths
• the prism CLI surface
• Hydra config keys
• checkpoint keys
• launcher flags
• None of the above

--verify is a new read-only flag on an operator script; no existing
invocation changes behavior.

How Has This Been Tested?

tools/ci/configure_branch_protection.py --verify   match -> exit 0
                                                   mismatch -> exit 1 (both confirmed)
ruff 0.16.9 over src tests tools scripts applications examples   clean
tools/ci/check_doc_links.py                                      495 links resolve
pytest tests/platform/test_branch_protection_contract.py
       tests/test_docs_consistency.py                            25 passed

• Platform tested on: Aurora login node, frameworks/2026.1.0.
• Tests skipped, and why: none relevant; this touches docs, one comment and
  one operator script.

New Dependencies, Data, or Assets

None.

Checklist

• I have performed a self-review of my own changes
• The code follows the project's style guidelines
• I have commented my code where the intent is not obvious
• I have made corresponding changes to the documentation
• I have added tests that cover this change, or explained why none are needed
• New and existing tests pass locally
• No secrets, credentials, or absolute institutional paths are included

--verify has no unit test: it only reads a live GitHub endpoint, so a test
would either stub the call it exists to make or require network and a token in
CI. Both exit paths were exercised by hand against the real repositories.

AI/LLM Assistance

• AI/LLM tools were primarily used to generate code or artifacts in this PR
• I have reviewed all such output line-by-line and understand it

Unticked pending a human read.

Additional Context

Not addressed here, and why: the SECURITY.md supported-versions table and the
two CHANGELOG promotions are all gated on the v0.1.0 tag, which NOTICE
blocks until the copyright holder is settled with Argonne tech transfer and the
other contributing institutions. Those are one commit once that question is
answered.