All three of those tweets are dumb.
1. If this were the case we wouldn't have seen Democratic losses in Hispanic areas in cities and we would have seen more substantial losses in the rural black belt. No reason to think people are suddenly going to start primarily identifying as rural in the absence of evidence.
2. I mean, they might, 2012 -> 2016 showed us that sometimes a candidate really can have a pretty unique coalition. I think it's probably more likely than not that Miami-Dade keeps sliding right but you never know.
3. 90% -> 93%. Need I say more about the Lincoln Project?
For the last time, four years of Lincoln Project Republicans ceasing to identify as Republicans entirely makes that last number meaningless.