How well do large language model-based chatbots perform in oral and maxillofacial radiology?