Assessing the Performance of Large Language Models on the Foreign Medical Graduate Examination (FMGE): Insights from GPT-4 Turbo, Gemini Advanced, and LLaMA 3.1 (70B)