The other three points don't strike me as significant evidence (1 cites "several colleagues" any of whom could have been the one, 3 strikes me as indeterminate

The other three points don't strike me as significant evidence (1 cites "several colleagues" any of whom could have been the one, 3 strikes me as indeterminate between someone being vague about attribution and your reading, 4 seems like guilty until proven innocent), but 2 is interesting. There is some gap between approving of the paper being pulled, and advocating it, but the comment makes the story more plausible. I am inclined to believe that Wilkinson is either making a bald-faced lie or telling the truth. I don't have grounds to accuse her of a direct lie, so I lean towards telling the truth. Because revoking acceptance is serious, I do not think it makes sense to suggest it was just a misunderstanding. Analogy: I'm invited to a party, and the host discusses inviting someone I'm not on great terms with. I'm noncommittal. They might guess that they should not invite that person to make me happy, because that's a relatively small step. However, it's unlikely they'll invite that person and then assault them on my behalf, because that would be an extreme measure--merely being noncommittal wouldn't suggest that. I think revoking acceptance is less severe than assaulting someone, but much much more severe than failing to invite them to a party. I believe a journal editor would only revoke an acceptance either 1) in the face of direct requests that they do so, or 2) as their own decision based on what they thought was necessary.