On P300 measurement stability: habituation, intra-trial block variation, and ultradian rhythms