On one level, I think these polls fruitfully remind us that Trump is likely to be quite competitive in most battleground states. As has been widely noted, Biden, the candidate who runs strongest against Trump, has slender registered voter leads in these polls of 3 points or less in the key Rustbelt battlegrounds of Michigan, Pennsylvania and Wisconsin. (Oddly,Biden runs stronger in these polls in Arizona where they give him a 5 point lead.)
They also confirm that, at this point, Biden does run the strongest against Trump in these states and that the differential between Biden's slim leads and the performance of other candidates like Warren and Sanders, while small, is enough to tip some of these states back in Trump's direction. The significance of this differential has been cloaked by polling that has shown Biden farther ahead in these states, so that lagging his performance by a few points was not enough to tip the states in Trump's direction.
That said, I do wonder about some of these results. Again, look at the three Rustbelt battlegrounds of Michigan, Pennsylvania and Wisconsin: previous polling had Biden ahead of Trump by an average of 11 points in Michigan, 10 points in Pennsylvania and 7 points in Wisconsin. And this is in a national environment where Biden leads Trump by an average of 9 points. So the NYT/Sienna results are pretty different. (Note: the results shown below for MI include the NYT//Siena likely voter result, but this doesn't really affect the comparison since the LV and RV results barely differed.)
That's not to say they're wrong. It could be the previous polls that were wrong. At any rate, if you scrutinize the NYT/Siena methodology document, it's easy to see ways in which their approach could have introduced error--or corrected it! Impossible to tell.
So the safest thing is to treat this set of polls as a new data point, but not a definitive one. As always, it's best to treat a single poll's findings in the context of data and trend from other sources.