fix: 飞书长连接 Pipeline 预初始化 + 错误日志强制刷新

- FeishuLongConnService.__init__ 预初始化 Pipeline（避免首条消息懒加载超时）
- 初始化失败时打印完整堆栈，启动日志里就能看到错误
- _process_message_safe 异常时 flush stderr 确保日志不被缓冲
- 消息处理时优先用预初始化的 pipeline，fallback 到 service_manager

src/integrations/feishu_longconn_service.py

@@ -55,6 +55,14 @@ class FeishuLongConnService:
         self._executor = ThreadPoolExecutor(max_workers=5, thread_name_prefix="feishu_msg")
         self._msg_count = 0
 
+        # 预初始化 Pipeline（避免首条消息时懒加载导致超时）
+        try:
+            self._pipeline = service_manager.get_pipeline()
+            logger.info("✅ MessagePipeline 预初始化成功")
+        except Exception as e:
+            logger.error(f"❌ MessagePipeline 预初始化失败: {e}", exc_info=True)
+            self._pipeline = None
+
     def _handle_message(self, data: P2ImMessageReceiveV1) -> None:
         """收到消息事件，立即提交到线程池，确保 SDK 回调快速返回"""
         self._msg_count += 1
@@ -68,6 +76,8 @@ class FeishuLongConnService:
             self._process_message(data)
         except Exception as e:
             logger.error(f"❌ 消息 #{msg_no} 处理异常: {e}", exc_info=True)
+            import sys
+            sys.stderr.flush()
             try:
                 mid = data.event.message.message_id
                 self._reply_message(mid, "抱歉，系统遇到了一些问题，请稍后重试。")
@@ -166,8 +176,9 @@ class FeishuLongConnService:
             # 使用 Pipeline 统一处理消息
             session_user_id = f"feishu_{chat_id}_{sender_id}"
             try:
-                pipeline = service_manager.get_pipeline()
-                logger.info(f"Pipeline 开始处理: user={session_user_id}, chat_id={chat_id}")
+                logger.info(f"正在获取 Pipeline...")
+                pipeline = self._pipeline or service_manager.get_pipeline()
+                logger.info(f"Pipeline 开始处理: user={session_user_id}")
                 response_data = pipeline.handle_message(
                     user_id=session_user_id,
                     message=text_content,