feat: /qe-fleet run 安徽运八需求 + 系统坑修复

系统修复:
- fleet_runner.py: Windows UTF-8 编码兼容（reconfigure stdout/stderr）
- fleet_runner.py: run 命令自动生成合并清单供导出使用
- fleet_manifest.py: 新增 save_merged_manifest() 保存统一清单
- fleet_config.yml: 修复 YAML keywords 格式（block sequence → flow sequence）
- 恢复被误删的知识库文件(payment_flow_cases, marketing_activity_cases)

QE Fleet 产物 (安徽运八需求):
- 153个测试点 + 53条可执行测试用例
- 需求分析/风险评估/测试策略/质量裁决（PASS）
- Excel 导出 + 版本快照 v2
- 确认结论文件已归档

output/manifests/安徽运八需求_prepare.json

@@ -0,0 +1,43 @@
+{
+  "base_name": "安徽运八需求",
+  "requirement_source_file": "E:\\test\\QaAutomationHub\\source_docs\\requirements_raw\\安徽运八需求.docx",
+  "requirement_input_type": "docx",
+  "normalized_requirement_file": "E:\\test\\QaAutomationHub\\output\\normalized_inputs\\安徽运八需求\\requirement.md",
+  "normalized_dir": "E:\\test\\QaAutomationHub\\output\\normalized_inputs\\安徽运八需求",
+  "technical_solution_files": [],
+  "normalized_technical_solution_files": [],
+  "project_profile_file": "E:\\test\\QaAutomationHub\\knowledge_base\\00_project\\project_profile.md",
+  "document_confidence": {
+    "requirement": 0.9,
+    "issues": []
+  },
+  "activated_knowledge": {
+    "terminology": {
+      "permanent": [
+        "E:\\test\\QaAutomationHub\\knowledge_base\\01_standards\\terminology.md"
+      ],
+      "optional": []
+    },
+    "semantic_matches": []
+  },
+  "knowledge_gaps": [
+    {
+      "category": "history",
+      "suggestion": "未激活任何历史缺陷/易漏场景，建议补充相关知识库条目"
+    },
+    {
+      "category": "best_practice",
+      "suggestion": "未激活任何最佳实践范例，建议补充同类型需求的优秀用例"
+    }
+  ],
+  "agent_notes": {
+    "document-parser": "解析完成，置信度 90%",
+    "knowledge-activator": "激活 1 常驻 + 0 可选术语"
+  },
+  "_meta": {
+    "zone": "prepare",
+    "base_name": "安徽运八需求",
+    "updated_at": "2026-07-13T01:33:01.474632+00:00",
+    "status": "completed"
+  }
+}