Are point mutations or DNA rearrangements responsible for the restriction fragment length polymorphisms that are used to type bacteria?