Their notification was fine. The tweet was vile and full of expletives. You'll have trouble finding any business or institution that tolerates speech like that. This isn't the censorship hill anyone should want to die on.
I do think that going back a decade with an algorithm and getting people to clean up tweets that likely match a regex filter is whitewashing and it does make twitter a less valuable historical resource. It's a totally different argument from the normal you should be able to say what you want - because I would support censorship of such tweets in the present.
If they want to enforce a policy consistently and not have to face constant complaints, then I don't think grandfathering in old violations after some arbitrary date is going to work.