AIRR export: use cdr3, junction or sequence fields depending on region_type
import & random data generation: ensure region_type is always stored in the repertoires
repertoire: get region type for all sequences in the repertoire
Kmer/PositionHelper: handle IMGT positions for CDR3 and junction when sequence_type is amino_acid. For nucleotide sequences, IMGT positions are not defined and throws an explicit error (before it didnt explicitly handle the nt case)
Kmer encoder: when retrieving the sequence_type from the model, default to EnvironmentSettings.sequence_type instead of None
(the 'added reference encoders as option for sklearn methods' is a lingering commit from before which is already on master)
(the 'added reference encoders as option for sklearn methods' is a lingering commit from before which is already on master)