docs: add N’KEMBA institutional reconstruction cross-walk

[riscas028]

Adds a non-normative downstream-consumer cross-walk for N’KEMBA, following the venue and scope guidance in #274.

The cross-walk:

• leaves TRACE v0.2 schema, wire format, claims, verification rules and conformance requirements unchanged;
• treats TRACE evidence as upstream evidence rather than institutional proof;
• preserves the §3.1.2 references boundary: a reference is a pointer, not attested truth about its target;
• separates runtime/session evidence, action evidence, human authority, operative document/version, external outcome, and institutional adoption;
• states both what the mapping covers and what N’KEMBA does not claim;
• adds the corresponding Crosswalks navigation entry in mkdocs.yml.

The N’KEMBA demonstrator mentioned in the document is explicitly described as a synthetic fail-closed functional-logic demonstrator, not TRACE conformance, certification, endorsement, or independent audit. The v0.7 default-pass issue discussed in agentrust-io/trace-tests#96 was corrected before this contribution was prepared.

Related: #223, #274; agentrust-io/trace-tests#96.

[riscas028]

Maintainer-facing note following the latest verifier/reference discussions: this cross-walk intentionally treats two boundaries as downstream hand-off points rather than TRACE claims. First, an unresolved or insufficiently established external fact remains explicitly unproved downstream; it is not promoted to success or converted into TRACE verification failure. Second, resolving a references URI does not by itself establish object identity, content immutability, authority, or institutional status; where a digest/binding is available N’KEMBA preserves it, and where it is absent the downstream conclusion remains bounded accordingly. This is consistent with the PR’s existing non-normative scope and does not propose a TRACE schema change. If maintainers prefer, I can fold this clarification into the cross-walk text during review rather than add normative language.

[rajnisht7]

@riscas028 The “Public reproducibility surface” section links to https://nkemba.pt/nkemba-evidence-demo-v08.html, but this currently returns 404 Not Found, so the documented public reproducibility surface is broken. The referenced demo-evaluator-v08.js is still reachable, but that doesn't make the HTML demo link valid.

[riscas028]

Fixed — thank you. I removed the unavailable HTML demonstrator URL from the public reproducibility section and left only the currently reachable evaluator source. The text now explicitly notes that the previous HTML URL is not currently available.

[imran-siddique]

Review/test follow-up on 675419d48231.

CI and CodeQL both ran and passed after release. The assurance boundary is stated correctly, but this overlaps #291 at the same document and navigation entry. Please consolidate on #291. If this version remains the candidate, align the "current public functional demonstrator" wording with the later disclosure that the HTML demonstrator is unavailable; describe the published evaluator source precisely.

Full local python -m pytest -q (Windows, Python 3.12, existing review environment): 5 failed, 1162 passed, 1 skipped in 11.39s.

The five local failures are the four generated-fixture byte comparisons and test_every_schema_in_the_repository_is_classified, matching the Windows newline/path-separator failures seen across these branches. The hosted Linux suite is green; the local suite is not.

Contribution requirement still outstanding: these commits have no Signed-off-by: trailer: 77e38062, 742bbfd1, 21df30d6, 675419d4. Please have the respective commit authors add their own DCO sign-offs, including merge commits, then request review of the resulting head.

Fresh workflow evidence: 34042337246 (success), 34042337370 (success).