langchain_nvidia_ai_endpoints[patch]: Invoke callback prior to yieldi…

…ng token (#18271)

## PR title
langchain_nvidia_ai_endpoints[patch]: Invoke callback prior to yielding

## PR message
**Description:** Invoke callback prior to yielding token in _stream and
_astream methods for nvidia_ai_endpoints.
**Issue:** #16913
**Dependencies:** None

libs/partners/nvidia-ai-endpoints/langchain_nvidia_ai_endpoints/chat_models.py

@@ -161,9 +161,9 @@ def _stream(
         inputs = self.custom_preprocess(messages)
         for response in self.get_stream(inputs=inputs, stop=stop, **kwargs):
             chunk = self._get_filled_chunk(self.custom_postprocess(response))
-            yield chunk
             if run_manager:
                 run_manager.on_llm_new_token(chunk.text, chunk=chunk)
+            yield chunk
 
     async def _astream(
         self,
@@ -175,9 +175,9 @@ async def _astream(
         inputs = self.custom_preprocess(messages)
         async for response in self.get_astream(inputs=inputs, stop=stop, **kwargs):
             chunk = self._get_filled_chunk(self.custom_postprocess(response))
-            yield chunk
             if run_manager:
                 await run_manager.on_llm_new_token(chunk.text, chunk=chunk)
+            yield chunk
 
     def custom_preprocess(
         self, msg_list: Sequence[BaseMessage]