If "convergence delay" is meant to refer to a delay that occurs before *relays* agree on the same set of revocations, then that term carries a presupposition that relays *will* at some point converge. As far as I can see, there isn't any particularly strong reason to think that they will ever conver...