A reconsideration of the Campylobacter dose–response relation