Validity and limitations of a bimodal parametric approach for the evaluation of antibody affinity distributions