Gestational Diabetes validation data study: an example of testing differences in sensitivity and specificity of two diagnostic methods using non-random sampling in a paired study design