Comment on "Semiquantum-key distribution using less than four quantum states"