[codex] Harden thulr eval telemetry and delegation prompts

[justintime109]

Summary

• expands the thulr eval bridge with richer trace metadata, trace inspection, failure labels, eval-set/redaction/rate/judge-bin support, efficiency guardrails, configurable noise bands, and numeric gate delta summaries
• adds calibration canaries, filters them out of the release gate candidate, updates the committed thulr baseline, and changes the fixture key from MAGIC_TOKEN to SAMPLE_IDENTIFIER to avoid secret-shaped retrieval noise
• improves same-agent vote runs with complementary voter stances and gives orchestrate workers the overall goal/contract alongside their assigned subtask
• updates README/reference docs, changelog, tests, and the thulr eval audit ledger

Validation

• npm ci
• npm run check
• npm run scan:privacy:staged

[justintime109]

@codex please review this PR.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 5881b32f85

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".

[justintime109]

@greptile review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 9815174ae6

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".