v0.5.5 - Allow more detail in AI failure alerts (#926)

🌟 Summary

v0.5.5 makes AI-generated GitHub Actions failure alerts more informative, especially for complex failures that need evidence and clear next steps.

📊 Key Changes

• 🧠 Expanded failure-triage detail: Alerts can now use up to 150 words and four bullets, with no line limit. The prompt encourages the AI to include decisive evidence, a likely cause, a fix or next diagnostic check, and other distinct failures.
• 🔎 Clearer confidence: The prompt asks the AI to distinguish confirmed facts from hypotheses and say when logs don’t establish a cause. Simple failures can still receive brief alerts.
• 📘 Updated documentation and version: The Failure Alert Action docs now describe the expanded triage, and the package version is bumped to 0.5.5.
• ✅ Validation: All 179 existing tests passed, and the changed Python files passed the repository’s Ruff checks and formatting.

🎯 Purpose & Impact

• 🛠️ Helps maintainers understand why a workflow failed and what to try next, reducing guesswork during troubleshooting.
• 📬 More detailed alerts may be especially useful when multiple jobs fail or logs contain important clues, while keeping straightforward failures concise.
• 🔒 The change only adjusts the AI prompt: model selection, log collection, Slack delivery, and fallback behavior remain unchanged.

What's Changed

• Allow more detail in AI failure alerts by @glenn-jocher in #926

Full Changelog: v0.5.4...v0.5.5