feat(chat): 基于 RAG 的笔记内容 AI 问答功能

实现类似 Google NotebookLM 的效果：笔记生成后自动向量化，
用户可针对笔记内容进行 LLM 问答。

### 后端
- 新增 VectorStoreManager（ChromaDB），按标题/转录分块建立向量索引
- 新增 chat_service.py RAG 问答：检索相关片段 → 构建 prompt → 调用 LLM
- 新增 /chat/index, /chat/ask, /chat/status API 端点
- 笔记生成完成后自动建立向量索引

### 前端
- 使用 @ant-design/x Bubble.List + Sender 组件构建聊天面板
- 新增 chatStore（Zustand + persist）持久化聊天记录
- MarkdownViewer 右侧嵌入 ChatPanel，通过"AI 问答"按钮切换
- 首次打开自动检查/触发索引，支持重新索引

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

backend/app/routers/chat.py

@@ -0,0 +1,74 @@
+from typing import Optional
+
+from fastapi import APIRouter
+from pydantic import BaseModel
+
+from app.services.chat_service import chat as chat_service
+from app.services.vector_store import VectorStoreManager
+from app.utils.logger import get_logger
+from app.utils.response import ResponseWrapper as R
+
+logger = get_logger(__name__)
+
+router = APIRouter()
+
+
+class IndexRequest(BaseModel):
+    task_id: str
+
+
+class ChatMessage(BaseModel):
+    role: str
+    content: str
+
+
+class AskRequest(BaseModel):
+    task_id: str
+    question: str
+    history: list[ChatMessage] = []
+    provider_id: str
+    model_name: str
+
+
+@router.post("/chat/index")
+def index_task(data: IndexRequest):
+    """为笔记建立向量索引。"""
+    try:
+        store = VectorStoreManager()
+        store.index_task(data.task_id)
+        return R.success(msg="索引完成")
+    except Exception as e:
+        logger.error(f"索引失败: {e}")
+        return R.error(msg=f"索引失败: {str(e)}")
+
+
+@router.get("/chat/status")
+def chat_status(task_id: str):
+    """检查笔记是否已建立向量索引。"""
+    try:
+        store = VectorStoreManager()
+        indexed = store.is_indexed(task_id)
+        return R.success(data={"indexed": indexed})
+    except Exception as e:
+        logger.error(f"查询索引状态失败: {e}")
+        return R.success(data={"indexed": False})
+
+
+@router.post("/chat/ask")
+def ask_question(data: AskRequest):
+    """基于笔记内容的 RAG 问答。"""
+    try:
+        history = [{"role": m.role, "content": m.content} for m in data.history]
+        result = chat_service(
+            task_id=data.task_id,
+            question=data.question,
+            history=history,
+            provider_id=data.provider_id,
+            model_name=data.model_name,
+        )
+        return R.success(data=result)
+    except ValueError as e:
+        return R.error(msg=str(e))
+    except Exception as e:
+        logger.error(f"Chat 问答失败: {e}", exc_info=True)
+        return R.error(msg=f"问答失败: {str(e)}")