We consider only those inputs that are not NULL. We then use `intersect` on pairs of inputs' rownames to identify those subjects are shared among all non-NULL inputs.
make_id2keep(probs, pheno, addcovar = NULL, kinship = NULL)
an allele probabilities array
a phenotypes matrix
a covariates matrix
a kinship matrix
a character vector of subject IDs common to all (non-null) inputs