fix(stream): 跳过空 choices 的 SSE 帧并记录 usage，避免流式解析越界

src/llm_models/model_client/openai_client.py

@@ -270,7 +270,15 @@ async def _default_stream_response_handler(
             # 如果中断量被设置，则抛出ReqAbortException
             _insure_buffer_closed()
             raise ReqAbortException("请求被外部信号中断")
-
+        # 空 choices / usage-only 帧的防御
+        if not getattr(event, "choices", None) or len(event.choices) == 0:
+            if getattr(event, "usage", None):
+                _usage_record = (
+                    event.usage.prompt_tokens or 0,
+                    event.usage.completion_tokens or 0,
+                    event.usage.total_tokens or 0,
+                )
+            continue  # 跳过本帧，避免访问 choices[0]        
         delta = event.choices[0].delta  # 获取当前块的delta内容
 
         if hasattr(delta, "reasoning_content") and delta.reasoning_content:  # type: ignore