The consistency of consistent histories: A reply to d'Espagnat