feat: 重构知识库系统，移除Hermes集成，增强RAG和同步功能

主要变更:
- 移除Hermes智能体及相关回调服务
- 新增知识库RAG、同步、调度、规范化和索引任务服务
- 重构orchestrator服务，增强运行时聊天功能
- 更新前端聊天、政策制度、设置等页面样式和逻辑
- 更新expense_claims和document_intelligence服务
- 删除llm_wiki相关服务和测试文件
- 更新docker-compose配置和启动脚本

server/tests/test_ontology_service.py

@@ -310,6 +310,44 @@ def test_semantic_ontology_service_keeps_travel_amount_follow_up_in_knowledge_qu
         assert result.clarification_required is False
 
 
+def test_semantic_ontology_service_rejects_draft_intent_inside_knowledge_session(monkeypatch) -> None:
+    session_factory = build_session_factory()
+    with session_factory() as db:
+        service = SemanticOntologyService(db)
+        monkeypatch.setattr(
+            service,
+            "_parse_with_model",
+            lambda **kwargs: LlmOntologyParseResult(
+                scenario="expense",
+                intent="draft",
+                confidence=0.91,
+                clarification_required=True,
+                clarification_question="请补充招待对象和票据附件。",
+                missing_slots=["participants", "attachments"],
+                ambiguity=[],
+                entity_hints=[],
+            ),
+        )
+
+        result = service.parse(
+            OntologyParseRequest(
+                query="我要去北京出差3天，一共可以报销多少钱？",
+                user_id="pytest",
+                context_json={
+                    "role_codes": ["employee"],
+                    "name": "曹笑竹",
+                    "grade": "P3",
+                    "session_type": "knowledge",
+                },
+            )
+        )
+
+        assert result.scenario == "knowledge"
+        assert result.intent == "query"
+        assert result.clarification_required is False
+        assert result.clarification_question is None
+
+
 def test_semantic_ontology_service_prefers_expense_for_customer_entertainment_narrative() -> None:
     session_factory = build_session_factory()
     with session_factory() as db: