patch: remove usage of llm, chat model __call__ (#20788)

- `llm(prompt)` -> `llm.invoke(prompt)`
- `llm(prompt=prompt` -> `llm.invoke(prompt)` (same with `messages=`)
- `llm(prompt, callbacks=callbacks)` -> `llm.invoke(prompt,
config={"callbacks": callbacks})`
- `llm(prompt, **kwargs)` -> `llm.invoke(prompt, **kwargs)`

libs/community/langchain_community/llms/ctransformers.py

@@ -97,7 +97,7 @@ class CTransformers(LLM):
         Example:
             .. code-block:: python
 
-                response = llm("Tell me a joke.")
+                response = llm.invoke("Tell me a joke.")
         """
         text = []
         _run_manager = run_manager or CallbackManagerForLLMRun.get_noop_manager()
@@ -125,7 +125,7 @@ class CTransformers(LLM):
 
         Example:
             .. code-block:: python
-                response = llm("Once upon a time, ")
+                response = llm.invoke("Once upon a time, ")
         """
         text_callback = None
         if run_manager: