The way it was implemented was that a user was basically two accounts - one voted, one commented. You could ban either one without knowing the other. The point being that if the issue was vote manipulation, you could ban the voting agent and be done with it. If the issue was as content violation then you could ban the other account and be done with it. It was literally just like having an app where you can log in with two accounts at once and choose which one to use to vote vs comment.