DINO-CVA: A Multimodal Goal-Conditioned Vision-to-Action Model for Autonomous Catheter Navigation