Fix vlm engine changes in mcore (#15076)

Author: meatybobby

nemo/collections/vlm/inference/vlm_engine.py

@@ -38,7 +38,7 @@ def generate(
         for i in range(len(prompts)):
             prompt = prompts[i]
             image = images[i] if images is not None else None
-            prompt_tokens, image_dict = self.text_generation_controller.tokenize_prompt(prompt, image)
+            prompt_tokens, image_dict = self.controller.tokenize_prompt(prompt, image)
 
             # Reuse encoder_prompt from scheduler to pass image
             request_id = self.scheduler.add_request(