fix(ci): fail archived broken live links

[konard]

Summary

• fail broken-link validation whenever Lychee reports a broken live URL
• retain Web Archive checks as actionable replacement diagnostics only
• run the terminal explanation with always() so helper failures cannot hide Lychee failures
• add a workflow-policy regression test and changelog fragment

Reproduction

Before this fix, a Lychee report containing a dead URL with an available Wayback snapshot caused check_web_archive.py to emit all_archived=true. The final workflow condition required both a nonzero Lychee exit and all_archived != true, so the failure step was skipped and CI passed while the source still referenced the dead live URL.

The regression test was added first and failed against the old workflow because the required unconditional failure step did not exist.

Verification

• pytest tests/test_workflows.py tests/test_check_web_archive.py -q
• ruff check .
• ruff format --check .
• mypy src/
• python scripts/check_file_size.py
• pytest --cov=src --cov-report=term (71 passed, 100% package coverage)
• sphinx-build -W --keep-going -b html docs /tmp/issue54-docs-build

Fixes #54

[konard]

Working session summary

Implemented and finalized the fix in PR #57.

• Broken live links now fail CI even when Wayback snapshots exist.
• Archive checks remain as replacement diagnostics.
• Added a regression test and changelog fragment.
• Local suite passes: 71 tests, Ruff, mypy, file-size checks, and Sphinx.
• Fresh Broken Link Checker, Security, and CI/CD workflows all pass.
• Branch is clean, current with main, and PR is ready for review.

---

This summary was automatically extracted from the AI working session output.

[konard]

🤖 Solution Draft Log

This log file contains the complete execution trace of the AI solution draft process.

💰 Cost estimation:

• Model: GPT-5.6 Sol
• Provider: OpenAI
• Public pricing estimate: $2.198127

📊 Context and tokens usage:

• 68.1K / 200K (34%) input tokens, 6.6K / 128K (5%) output tokens

Total: (68.1K + 1.2M cached) input tokens, 6.6K output tokens, $2.198127 cost

🤖 Models used:

• Tool: OpenAI Codex
• Requested: gpt-5.6-sol
• Thinking level: low (~7999 tokens)
• Model: GPT 5.6 Sol (gpt-5.6-sol)

📎 Log file uploaded as Gist (908KB)

• View complete solution draft log

---

Now working session is ended, feel free to review and add any feedback on the solution draft.

[konard]

🎉 Auto-merged

This pull request has been automatically merged by hive-mind.

• All CI checks have passed

---

Auto-merged by hive-mind with --auto-merge flag