Enhance troubleshooting analyzers: select/release inference, detailed reports, and error classification

[mouxinqq]

Motivation

• Improve accuracy of select/release correlation and worker-type attribution for better load and cache diagnostics.
• Provide richer, cross-linked detail reports to make triage of cache, load, health, latency and trace issues easier.
• Surface error impact and related URLs to speed up root-cause identification.

Description

• Add inference logic in log_parser.match_select_release to infer worker type for ambiguous release and token-release events and emit unmatched_releases, type_summary, and id_coverage fields.
• Add request-count select/release regex and last-counter-state analysis (_analyze_counter_last_state) to analyzers/load.py and surface counter_last_state in reports.
• Add cross-diagnosis for cache (_analyze_cross_diagnosis) and include detailed detail/cache_* outputs; format_cache_report now returns both summary and detail strings.
• Improve health analyzer by including REGISTERED events, computing detection latency (_compute_detect_latency), and exposing 检测延迟 in worker table and detail file.
• Enhance errors analyzer: add SOURCE_LAYER_RULES/IMPACT_RULES entry for missing YAML, add classify_impact, extract related URLs with _extract_urls, and include impact/URLs in error_top_n.
• Trace analyzer now calls match_select_release to include select/release checks in trace diagnoses and report sr_check.
• Update various report formatters (load_report.py, latency.py, chart.py, load_report.format_load_report) to link to newly created detail/ files and to show clarifying metric notes.
• Orchestrator troubleshoot.py now aggregates and writes many additional detail files (cache, load, latency, counters, traces), improves status summary dedup/length, and documents status definitions.
• Minor table rendering tweak in chart.py and added CLI self-test case for token-release type inference in log_parser.py.

Testing

• Ran python3 .claude/skills/troubleshoot/scripts/log_parser.py self-test which includes a token-release inference sample; the added checks passed.
• Executed troubleshoot.py on a small synthetic log sample to verify report generation and creation of summary/troubleshoot_report.md and corresponding detail/ files; report generation succeeded.
• No other automated unit tests were modified; existing script-level self-tests and report generation checks passed.

---

Codex Task