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/anyscale.py

@@ -75,7 +75,7 @@ class Anyscale(BaseOpenAI):
             # To leverage Ray for parallel processing
             @ray.remote(num_cpus=1)
             def send_query(llm, text):
-                resp = llm(text)
+                resp = llm.invoke(text)
                 return resp
             futures = [send_query.remote(anyscalellm, text) for text in texts]
             results = ray.get(futures)