I don´t like the idea, too. Basically you would choose between two different random distributions with two different means , mean(X) and mean(Y).
If mean(X) > mean (Y), why would you ever choose Y ?
Special spins with 1 respin per slot, don´t give many choices for you. In ~ 1,1% of all cases you can choose the rare ((6*8/(4^6)) (case 4 rares in spin, cards different, not sure about real number of cards in spin). Estimations could soon follow.