Comment on ``Loss-error compensation in quantum-state measurements''