The point about the 3rd Congressional District moving further left is a very important one to remember, and is one of the key reasons Clinton was able to eke out a victory in the state, but another important and thus far unmentioned point is this: Evan McMullin was on the ballot in Minnesota, as opposed to the other typically-Democrat Midwestern states, and received 53,076 votes there (Clinton won the state by 44,765). Also, it's important to remember Trump came in 3rd in the Republican Primary in this state, and the state's Republican Party is historically more moderate than not.
It doesn't really work that way.
First of all, you'd have to assume that McMullin voters were uniformly likely right-wing
and likely Trump voters in the absence of said choice. While I'm sure a huge contingent of McMullin voters may have been default GOP voters in other circumstances, there's also quite a bit of evidence that shows third-party defections in the upper Midwest (centered around the Dakotas) were just as much fueled by Obama voters as Trump voters. It's arguably the one part of the country where the Left may have lost as many or more voters because of third-party choices (especially to Johnson, which wasn't mentioned in your comparison, but we'll exclude Johnson's effect for simplicity's sake).
Next, you'd have to assume that all (or practically all) of these voters would have came out to vote and cast votes in the presidential contest. Some would have stayed home and some would have just left the ballot blank. Maybe not that many, but it doesn't have to be many, as I'll show below.
Finally, Trump would have needed to win almost all of those 53k McMullin voters (
including any who stayed home or who left ballot blank) to flip the state.
If every single McMullin voter did in fact show up and did break for one of the two parties - with Trump winning 80% and Clinton getting 20% - Hillary still wins MN by 12,919 votes.
Using the above scenario but increasing Trump's vote share to 90% among them still leads to Hillary winning by 2,305 votes.
If just 10% of McMullin voters stayed home/didn't vote and Trump won 95% of those who did vote, Clinton still wins by 1,773 votes. There's almost no reasonable scenario where McMullin voters would have flipped the state in favor of Trump when considering margins and when considering who/why they were voting for McMullin in the first place, and that's of course still ignoring that we could just as easily say "Clinton would have won anyway if you deduct favorable third-party votes from other candidates and give them to her".
