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/tests/integration_tests/llms/test_tongyi.py

@@ -8,7 +8,7 @@ from langchain_community.llms.tongyi import Tongyi
 def test_tongyi_call() -> None:
     """Test valid call to tongyi."""
     llm = Tongyi()
-    output = llm("who are you")
+    output = llm.invoke("who are you")
     assert isinstance(output, str)