refactor: simplify thinking process UI and reduce noise

- Remove redundant status steps (understanding, preparing, generating)
  from SSE stream — only emit retrieved docs and reasoning content
- ThinkingProcess: only show when there's actual reasoning or docs
- Collapse header shows concise state: thinking count or doc count
- Clean up unused icon imports

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

backend/src/rag/chains.py

@@ -133,37 +133,34 @@ class RAGChain:
         """流式调用（返回答案流和文档，包含思考过程）"""
         import time
 
-        # 0. 思考阶段开始
         start_time = time.time()
-        yield {"type": "thinking", "stage": "understanding", "message": "正在理解问题..."}
 
         # 1. 检索文档
-        yield {"type": "thinking", "stage": "retrieving", "message": "正在检索相关知识..."}
         retrieval_start = time.time()
         docs = await self.retriever.ainvoke(question)
         retrieval_time = time.time() - retrieval_start
 
-        # 发送检索结果 — 包含文档标题和摘要
-        doc_details = []
-        for i, doc in enumerate(docs[:5]):
-            metadata = doc.metadata if hasattr(doc, 'metadata') else {}
-            title = metadata.get("title", metadata.get("filename", f"文档 {i+1}"))
-            preview = doc.page_content[:100].replace('\n', ' ')
-            doc_details.append(f"**{title}**: {preview}...")
+        # 发送检索结果（只在有文档时）
+        if docs:
+            doc_details = []
+            for i, doc in enumerate(docs[:5]):
+                metadata = doc.metadata if hasattr(doc, 'metadata') else {}
+                title = metadata.get("title", metadata.get("filename", f"文档 {i+1}"))
+                preview = doc.page_content[:100].replace('\n', ' ')
+                doc_details.append(f"**{title}**: {preview}...")
 
-        yield {
-            "type": "thinking",
-            "stage": "retrieved",
-            "message": f"检索到 {len(docs)} 篇相关文档",
-            "doc_count": len(docs),
-            "time": round(retrieval_time, 2),
-            "details": doc_details
-        }
+            yield {
+                "type": "thinking",
+                "stage": "retrieved",
+                "message": f"检索到 {len(docs)} 篇相关文档",
+                "doc_count": len(docs),
+                "time": round(retrieval_time, 2),
+                "details": doc_details
+            }
 
         context = "\n\n".join(doc.page_content for doc in docs)
 
-        # 2. 构建prompt
-        yield {"type": "thinking", "stage": "generating", "message": f"基于 {len(docs)} 篇文档生成回答..."}
+        # 2. 构建prompt并流式生成
         prompt_value = await self.prompt.ainvoke({"context": context, "question": question})
         messages = prompt_value.to_messages()
 

backend/src/rag/conversation_chains.py

@@ -85,16 +85,8 @@ class ConversationChain:
         """流式调用（包含思考过程，捕获推理模型的真实推理内容）"""
         import time
 
-        # 思考阶段
         start_time = time.time()
-        yield {"type": "thinking", "stage": "understanding", "message": "正在理解问题..."}
-
-        # 准备历史
         history_messages = self._format_history(chat_history or [])
-        history_count = len([m for m in (chat_history or []) if m["role"] == "user"])
-        yield {"type": "thinking", "stage": "preparing", "message": f"加载对话上下文（{history_count} 轮历史）..." if history_count > 0 else "准备生成回答..."}
-
-        yield {"type": "thinking", "stage": "generating", "message": "正在生成回答..."}
 
         # 直接使用原始 OpenAI SDK 捕获推理内容
         prompt_value = await self.prompt.ainvoke({