The side effect of this function is to drop the rows (individuals)
with unclearly identified genotypes, i.e., with at least one allele
among na.alleles
.
Other variables in the data table are eventually checked and levels
with no observation (e.g., population) are dropped.