The article actually changed quite a bit over time.
From 7 minutes after the story was posted on Feb 21 11:17 am ET:
This version of the article is rather short and to the point. It seems to be written entirely by Zolan Kanno-Youngs, though headlines are usually written by editors.
The next version, about 9 hours later, with contributions by Erica L. Green, is significantly longer and is far more politically charged. This seems to be where the headline comes in:
The last online version seems to have appeared about 3 hours later with a much toned down headline:
They could have been A/B testing the headlines to see what drove traffic as the NYT is known to do. That would explain how the tweet’s version seems to have a date well after the above, but with the older headline. A/B testing seems to lead to charged headlines.
Their Feb 22 print edition doesn’t (and can’t) optimize headlines that way, so it’s more subdued:
This brings up a question. Is the print edition of the NYT better than the digital one? Sure, stories come out slower and they can’t update them, but at the same time they aren’t trying to drive online clicks or doing SEO.