>Isn't that the definition of rotation, the "exchanging one dimension for another"?

Maybe? I think it allows solutions not typically considered to be rotations, like the special relativity example.