I think we really need to consider the efficiency of the shuffle as well when evaluating a shuffling technique. If it takes hundreds (much less tens of thousands) of lengthy steps to shuffle the deck, then no one is going to do it.

Additionally, your analysis only considers when the permutation is reachable. You would also want to look at how likely each permutation is.