Your other Left! Vision-Language Models Fail to Identify Relative Positions in Medical Images-MICCAI 2025