use smolagents's step_callbacks instead of litellm's callback

app.py

@@ -23,19 +23,11 @@ DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
 rate_limiter = RateLimiter()
 
-class LiteLLMModelWithRateLimit(LiteLLMModel):
-    """API呼び出し前にレート制限をチェックするLiteLLMModelのラッパー"""
-    def run(self, **kwargs) -> dict:
-        # APIを呼び出す直前に待機処理を挟む
-        rate_limiter.wait_if_needed()
-        # 元のrunメソッドを呼び出してAPIアクセスを実行
-        return super().run(**kwargs)
-
 # --- Basic Agent Definition ---
 # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
 class BasicAgent:
     def __init__(self):
-        model = LiteLLMModelWithRateLimit(
+        model = LiteLLMModel(
             model_id="gemini/gemini-2.5-flash", 
             api_key=os.getenv("GOOGLE_API_KEY")
         )
@@ -48,6 +40,7 @@ class BasicAgent:
                 YoutubeTranscriptTool()
             ], 
             model=model,
+            step_callbacks=[rate_limiter.increment_and_sleep_if_needed]
         )
         print("BasicAgent initialized.")