To be a bit contrarian, I don't think an incumbent doing in the mid-40s against a few people who have not even unified their own party is all that great for him. The Dem numbers in this poll are basically irrelevant, all that matters is that Trump is within the same 1-2 points against all of them. Again, the candidates with the most name recognition among the Dems will benefit the most. The difference there is with the Hispanic and "Other" vote, who are the most marginalized and also the group undersampled by Siena in 2018.
That may explain some of their misses in House races too: In Texas, they had Gina Ortiz Jones down by 9 and she ended up losing by 0.5; they had Lizzie Fletcher down by 1 and 3, and she ended up winning by 5.
NYT/Siena's House polls were kind of a mixed bag - it seems like they fell into many of the same traps that other pollsters had wrt Hispanic constituencies and polling of those constituencies was pretty off-base. Overall, though, most of their polls were within the margin of error of the actual result, especially in the Midwest.
Also, going by that same logic, I think there is a case to be made that the Arizona numbers are the one piece of good news out of this whole slew of polls for the Democrats, and that Democratic strength in Arizona may be underestimated.