Fix:local variable 'response' referenced before assignment (#9230)

### What problem does this PR solve?

https://github.com/infiniflow/ragflow/issues/9227

### Type of change

- [x] Bug Fix (non-breaking change which fixes an issue)

---------

Co-authored-by: Kevin Hu <kevinhu.sh@gmail.com>

rag/llm/cv_model.py

     def chat_streamly(self, system, history, gen_conf, images=[]):
         from transformers import GenerationConfig
         ans = ""
+        response = None
         try:
             response = self.model.generate_content(
                 self._form_history(system, history, images),
         except Exception as e:
             yield ans + "\n**ERROR**: " + str(e)
 
-        yield response._chunks[-1].usage_metadata.total_token_count
-
+        if response and hasattr(response, "usage_metadata") and hasattr(response.usage_metadata, "total_token_count"):
+            yield response.usage_metadata.total_token_count
+        else:
+            yield 0
 
 class NvidiaCV(Base):
     _FACTORY_NAME = "NVIDIA"