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_yuan2.py

@@ -13,7 +13,7 @@ def test_yuan2_call_method() -> None:
         top_p=0.9,
         use_history=False,
     )
-    output = llm("写一段快速排序算法。")
+    output = llm.invoke("写一段快速排序算法。")
     assert isinstance(output, str)