Add reproducible Context Fabric text benchmark corpus

[Chris0Jeky]

Refs #2319

Context Fabric processors need shared, reproducible fixtures before their quality can be compared. Add the first CF-24A corpus slice: nine synthetic text/transcript fixtures, exact source hashes, reference outputs, licence/method/date metadata, and explicit hostile-injection controls.

The local command validates the corpus and its 16 KiB source-byte budget, then computes precision/recall/F1 by candidate kind when supplied predictions cover every fixture. Without predictions, metrics are unavailable. The example predictions are deliberately imperfect; no processor is executed or measured. Audio, image, PDF and runtime benchmark acceptance remains on #2319.

Validation at 6c9bec284b1feaade2cd49ce3bf9219c0b1d4320:

• py -3 -B -m unittest discover -s scripts/context_fabric -p "test_*.py": 21 passed.
• Corpus validation: 9 fixtures, 934 source bytes; example scoring and schema validation passed.
• node scripts/check-doc-links.mjs: 664 Markdown files, zero broken links.
• node scripts/check-docs-governance.mjs and git diff --check: passed.

The testing guide documents the command and its limits. Review-first execution is unchanged. Hosted CI and independent Terra review are separate pending gates. No live provider, external media, or real user data was used.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-08T01:09:12.815784Z	6c9bec2	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[Chris0Jeky]

Independent Terra review round 1 at 6c9bec284b1feaade2cd49ce3bf9219c0b1d4320 against db102dcba25cc5f347f22b38ce13d00502948ed9: no confirmed CRITICAL/HIGH findings.

The reviewer independently recomputed all nine source hashes and reference IDs (zero mismatches, 934 bytes), checked byte-preserving attributes, per-kind exact-set scoring, incomplete-prediction rejection, and unavailable metrics when no predictions are provided. No processor or media/runtime measurements are claimed.

One nonblocking documentation/validation parity finding is retained on #2319: the README describes validation against the JSON Schema, while the command uses manual validation and the schema test validates the schema itself. Numeric optional notes illustrates a schema/CLI difference; the committed corpus is unaffected. Disposition: tracked on the existing corpus issue for its next validation slice, no extra fix cascade in this PR.

Local 21-test, corpus, link and governance evidence remains as recorded. Hosted CI is a separate merge gate.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 6c9bec284b

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".