One problem with your numbers is that you did not mention that in 2016 the number of undecided voters was 25% and on a recent poll it was 10%. The probabilities favor that most of the 15% that are no longer undecided, have gone to the Democratic side. That means that the "small" percentage uptick in registered Republican voters is not enough to overcome that. In addition, what guarantees that the uptick in Republican voters is not an uptick that will vote against Trump and not for Trump. Scaramucci is a perfect example of a voter that is Republican but has stated he will not vote Democratic but will NOT vote for Trump. That uptick could be for that reason as Republican voters sign up in an attempt to vote Trump down versus another Republican candidate. trying to get the Republican party "back on track".
Keep in mind that there are now 2 parties within the Republican party with one of those being the Trump party and the other being the not-for-Trump party.
Gallup did show that in 2016 25% of all Americans disliked both Trump and Clinton, this included 54% of all independents.
One in Four Americans Dislike Both Presidential Candidates
Also according to CNN exit polls, 12% of all independents voted third party, against both Trump and Clinton. So you're right on there.
https://www.cnn.com/election/2016/results/exit-polls
Now if you go back to the generic presidential vote, question 68. You'll see 29% of independents say they'll vote for Trump, 30% for the Democratic candidate along with 20% stating it depends. I take that "It depends," crowd means their vote will be decided by whom the Democrats nominate. Nominate the right candidate, most of that 20% will vote Democratic, nominate the wrong candidate, they very well could vote third party again. Against both Trump and whomever is the Democratic nominee. My take on that anyway.
Now it's hard to realize that there are two parties within the GOP. Question 81A shows 68% of Republicans have a very favorable view of Trump vs. 9% who view him very unfavorable. The very favorable and very unfavorable are strong opinions or views of Trump. The somewhat's which I left out are wishy washy and can change easily. The very favorable's and very unfavorable's don't change much.
https://d25d2506sfb94s.cloudfront.net/cumulus_uploads/document/7jv630mjdk/econTabReport.pdf
Having 10% or so in the never Trump category is about the same percentage as in 2016 when 88% of Republicans voted for Trump, 8% for Clinton and 4% third party. That'a 12% adding up Clinton's and third party votes in 2016. The same percentage when you add up the very unfavorable and somewhat unfavorable together. Not much has changed in my opinion.
I expect Trump to get roughly the same 88% from Republicans that he received in 2016. Independents could be very different. Trump won them 46-42 over Hillary with 12% voting third party. We know those 12% don't like Trump. They'll probably either vote third party again if they don't like who the Democrats nominate or vote Democratic if they do. Anyway you look at it Trump is in trouble among independents. Go back to question 81A, 23% of independents view Trump very favorably, 38% very unfavorably. Again discounting the somewhat's which can change their view of him in a heartbeat. That's a minus 15 points among independents, a group he won by 4 points over Hillary. Not a good omen for Trump.
worst yet, go to question 99, run for reelection. 33% of independents want Trump to run for reelection, 46% do not. A minus 13 points there in a group he won in 2016 by 4 points. For me what all this is stating, it all boils down to whom the democrats nominate for 2020. The right candidate, the Democrats win in a landslide. The wrong candidate, another 2016 is a possibility. I wouldn't write that off. Not yet anyway.