Performance of large language models on the Korean Medical Licensing Examination and quality of explanation of gastroenterology questions: a comparative analysis of ChatGPT-3.5, ChatGPT-4.0 and Google Bard