Assessing the reliability of the borderline regression method as a standard setting procedure for objective structured clinical examination