Suggesting enhancement to account for stratification and multiple responses

@felixfan Is it possible to enhance PyHLA to include an option for conditional logistic regression? Some data could be derived from family members or matched donors. Also, what is a possible enhancement for dependent variables with multiple categories (control, disease1, disease2, ...etc) and in the same time clustering structure in the data. Finally, does PyHLA support haplotype analysis?