fix(inc-20260425): A1+A2 後續 — Solver/Critic timeout + auto_repair 接線 + Runbook + Grafana

延續 595629c0 INC-20260425 修復，補三段 Agent + 全鏈路觀測：

A1 後續 — Solver/Critic 三段 timeout 接線:
- solver_agent.py: AGENT_SOLVER_TIMEOUT_SEC=20.0（env override）
- critic_agent.py: AGENT_CRITIC_TIMEOUT_SEC=15.0（env override）
- protocol.py: 三 Agent 共用 observe_agent_step() 包裹呼叫
  · success/timeout/error outcome label
  · histogram 寫入 aiops_agent_step_duration_seconds

A2 後續 — auto_repair_service 改用 _diagnose_fallback_chain:
- auto_repair_service.py +46 行 — 切換 DIAGNOSE 路由到新 chain（NEMO→GEMINI→CLAUDE）
- 完全避開 Ollama CPU 238s 二次 timeout

新增 metrics:
- core/metrics.py +59 行 — 配合 observe_agent_step 的 histogram bucket + label cardinality

新增測試 (862 行):
- test_agent_step_timeouts.py (475) — 三 Agent 各 timeout 邊界 + outcome label
- test_ai_router_diagnose_fallback.py (387) — _diagnose_fallback_chain 正確序

新增配套:
- docs/runbooks/RUNBOOK-AGENT-STEP-LATENCY.md (350) — INC 故障排查 + 觀測指引
- ops/monitoring/grafana/agent_step_latency_rules.yaml (160)
  · 三 Agent histogram alert rules（p99 > timeout 80% → warning）

驗收: 33 tests pass (test_agent_step_timeouts 22 + test_ai_router_diagnose_fallback 11)

INC-20260425 雙修總工作量（595629c0 + 此 commit）:
  · 5 個 service/agent 檔修改
  · 1 個新 observability 模組
  · 4 個新測試/配套檔
  · 1372+187 = 1559 行新增

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 (INC-20260425 後續) <noreply@anthropic.com>

apps/api/src/services/auto_repair_service.py

@@ -500,6 +500,52 @@ class AutoRepairService:
                             playbook_id=playbook.playbook_id,
                             verification_result=verification_result,
                         )
+
+                        # 2026-04-27 P3.1-T1 by Claude — 三 Tier-1 服務整合
+                        # PostExecutionVerifier 判斷失敗/降級 → 觸發自動 Rollback
+                        if verification_result in ("failed", "degraded"):
+                            try:
+                                from src.services.rollback_manager import get_rollback_manager
+                                from src.services.declarative_remediation import DeclarativeRemediation
+                                from src.core.metrics import ROLLBACK_EXECUTED_TOTAL
+
+                                # 從 Incident 推導 target / namespace / action
+                                _rb_target = (incident.affected_services or ["unknown"])[0]
+                                _rb_ns = "awoooi-prod"
+                                _rb_action = f"kubectl rollout restart deployment/{_rb_target} -n {_rb_ns}"
+                                _spec = DeclarativeRemediation().evaluate(
+                                    action=_rb_action,
+                                    target=_rb_target,
+                                    namespace=_rb_ns,
+                                )
+                                rollback_mgr = get_rollback_manager()
+                                rollback_result = await rollback_mgr.trigger(
+                                    incident_id=incident.incident_id,
+                                    spec=_spec,
+                                    verification_result=verification_result,
+                                )
+                                _rb_status = "success" if rollback_result.success else "failed"
+                                _rb_reason = "converged" if rollback_result.convergence_confirmed else (
+                                    "no_previous_revision" if rollback_result.error and "revision" in (rollback_result.error or "")
+                                    else "error"
+                                )
+                                ROLLBACK_EXECUTED_TOTAL.labels(
+                                    status=_rb_status, reason=_rb_reason
+                                ).inc()
+                                logger.info(
+                                    "auto_rollback_triggered",
+                                    incident_id=incident.incident_id,
+                                    rollback_success=rollback_result.success,
+                                    convergence_confirmed=rollback_result.convergence_confirmed,
+                                    rollback_error=rollback_result.error,
+                                )
+                            except Exception as _rb_e:
+                                logger.exception(
+                                    "auto_rollback_failed",
+                                    incident_id=incident.incident_id,
+                                    error=str(_rb_e),
+                                )
+
                     except Exception as _inner_e:
                         logger.warning(
                             "auto_repair_verify_and_learn_failed",