A high rate of non-compliance confounds the study of whole grains and weight maintenance in a randomised intervention trial - the case for greater use of dietary biomarkers in nutrition intervention studies