Text this: Performance of large language models on veterinary undergraduate multiple-choice examinations: a comparative evaluation