Probing the Moral Development of Large Language Models through Defining Issues Test