I did a quick calculation a few weeks ago of an "all trends repeat" scenario (which obviously never happens, but whatever), though I didn't adjust to a national 50/50 map; Trump improves very slightly in the Electoral College, with ME/MN/NH/NV all flipping R while only AZ/UT flip D. That gives the following map:
Trump wins, 312-226, a slight improvement over 2016's 305-233 (discounting the Maine electoral vote).
However, it takes only a very slight universal swing to defeat Trump, while none of the Democrat's states are particularly close. Here are all victories with a margin of under 1%:
Nevada: Trump+0.08
Texas: Trump+0.46
Georgia: Trump +0.63
A swing of 0.46 is even less than real-life's necessary swing of 0.79. Flip Nevada and Texas and the Democrat wins, 270-268:
If you double the 2016 trends, rather than halving them, you get a narrow Democratic victory, 276-262:
From the previous map, Nevada, Rhode Island, and Georgia all flip.
My point being, while trends never necessarily continue in a certain direction (see Utah being the strongest-trending state two elections in a row; in 2012 towards the Republican, in 2016 towards the Democrat), the trends from 2016, if you assume consistent population distribution between states (which also never happens, but whatever) don't really point to an eternal Republican Electoral College lock; they point to neither party having much of a consistent advantage, which is in agreement with the historical data. If you plug in population trends, with Sun Belt states gaining electoral votes and Rust Belt states losing them, you come to the conclusion that there'll eventually be a narrow Democratic advantage to the Electoral College.
Except not -- because by then the coalitions will have shifted. As they inevitably do