See my reply further down. I disagree with that characterisation.
I also note that asking for particular changes was a mistake. They should have stuck with "take it down or we will take it down", to avoid people making the argument you just made.
If only because being seen to exercise editorial control might introduce significant new legal risks that the ToS, and its policing, is presumably meant to avoid in the first place.
If they have the final say on whether you can publish they have editorial control. A request to take down a whole article would be no better in my view. They are entitled to have TOS that do this and apply it strictly (although I don't actually see the breach in this case but we can we assume there was one for this discussion).
I don't like the article and if I was editing a paper or a website I would not publish the article in that form but DO is not a publisher but an infrastructure provider and for something that is not clearly illegal it seems highly inappropriate.
However they have guaranteed that I would not use them for any public facing service or content by stepping in this way (I might consider them for testing, internal or pure compute workloads).
I also note that asking for particular changes was a mistake. They should have stuck with "take it down or we will take it down", to avoid people making the argument you just made.
If only because being seen to exercise editorial control might introduce significant new legal risks that the ToS, and its policing, is presumably meant to avoid in the first place.