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/llamacpp.py

@@ -278,7 +278,7 @@ class LlamaCpp(LLM):
 
                 from langchain_community.llms import LlamaCpp
                 llm = LlamaCpp(model_path="/path/to/local/llama/model.bin")
-                llm("This is a prompt.")
+                llm.invoke("This is a prompt.")
         """
         if self.streaming:
             # If streaming is enabled, we use the stream