On the nonuniqueness of receiver function inversions