v0.2: Make DATA-01 decision-grade

[t3chn]

Closes #3.

Summary:

• Add three synthetic public DATA-01 cases with CSV and SQLite fixtures.
• Replace the sample DATA-01 scorer with config-driven deterministic checks for metrics.json, report.md, chart_spec.json, extra files, and critical caps.
• Align DATA-01 with the standard scorer-contract model: artifact_exact for required files, schema_contract for metrics.json/chart_spec.json, numeric_metric for exact/tolerance values, claim_rubric for report.md factual checks, and future security_leak checks for private canaries/honey rows.
• Add DATA-01 mutation support, sample artifact generation, data01-smoke, CI coverage, tests, and docs.

Test plan:

• make validate
• make test
• make smoke
• make compare-smoke
• make if01-smoke
• make data01-smoke
• make leak-check
• python3 -m ruff check .
• git diff --check

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 1db9638f39

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".

[chatgpt-codex-connector]

Enforce report word-limit failures as policy violations

When report_max_words fails, the scorer only records a failed check and does not add a policy violation or score cap, so overlong reports can still pass. For example, with otherwise-correct artifacts, a 500-word report in case_001 still returns success=True because the score remains above PASS_THRESHOLD and success is gated on not violations. This makes the configured max_words constraint effectively non-blocking and undermines the task’s “concise report” requirement.

Useful? React with 👍 / 👎.