If you assume D continues to be down 10% on in person voting and that also carries over to election day, and neutral on Absentee ballots, I'm estimating Clinton votes of 1.955 million (losing 10% of pre-vote and election day compared to Obama '12 = 210k lost votes [135k/74k]).
If you assume R continues to be down 6% in person and that carries over to election day and also continues down 15% on absentee ballots, I'm estimating Trump votes of 2.111 million (losing 6% of pre-vote and election day compared to Romney '12 = 127k lost votes [69k/58k] plus another 21k from absentee mail ins].
That would yield 48.1%C/51.9% Trump for NC. However, the incremental in-person votes among independents that is surging is heavily skewed towards whites and white men in larger than normal #s and Trump would probably pick up another 1% or so from that group so we are trending towards 47-53% if this continues. Of course, that is a big if - and third party votes could change things as well - but if I had to bet right now it looks like Trump takes NC on a bit lower turnout than 2012.
(if you assume election day voting = identical to 2012, it slightly moves it - 48.3%C/52.7%T)
I agree with the possibility of all of that. But I think some of the underperforming early vote may be due to polling station operating location and hours changes (for some counties) if there were any (reported in some main stream news headlines, and which I still would like to look at to understand). And so some of that “would-be early vote” could appear later on election day. That said, I don’t know the full details of those news articles yet.