为 Cron 失败写入 lastErrorReason，支持原因级运维

• Cron jobs already emit error logs and you can view job state fields.
• Team has a basic incident taxonomy (auth, timeout, quota, etc.).

1. Upgrade to a build containing PR #14382 and restart gateway.
2. Create controlled failure cases (expired key, forced timeout, invalid format) in staging jobs.
3. Inspect job state and verify lastErrorReason is populated with expected class.
4. Map each reason to runbook actions: retry/backoff/escalate/manual fix.
5. Add dashboard/alert routing by reason to reduce mean-time-to-diagnosis.

• This line is observability-first; scheduling policy changes are separate follow-ups（需验证）.
• Reason parsing may classify unknown provider errors as generic buckets.

This tip is aggregated from community/public sources and preserved with attribution.