Chasing the evidence: the influence of data segmentation on estimates of dynamic cerebral autoregulation