This is absolutely absurd.
Let’s say every single uncommitted superdelegate picks Bernie Sanders. That would give him 168 superdelegates compared to 547 for Hillary (as of the current “poll” that isn’t even a poll). That would be roughly 76%-24% in favor of Hillary. Now let’s double Nate Silver’s margin of error, so all the way up to 21.2% (152 delegates)! That leaves Hillary with 392 supers compared to 323 for Bernie… Wow! He *still* is losing in supers! In fact, the margin of error would have to be over 26% for Bernie to win a majority of supers. The margin of error to overtake her pledged delegate lead (and yes, she is incontrovertibly winning in pledged delegates) would need to be much much much larger. In fact, let’s say Bernie, despite inevitably getting killed in NJ, DC, and PR, is somehow able to narrow the pledged delegate gap to 100 delegates. That mean’s he’d need 408 supers, so your little “poll” would need over a 33% margin of error! That’s barely better than flipping a coin on any given delegate, and that’s after assuming every single uncommitted super is already supporting Bernie Sanders! Do you not see how absolutely impossible all of this is?
Sure, all of those events you outlined like a Biden/Warren ticket have a nonzero chance, but those events are converging to zero almost surely.