Add new residue to kmer on push instead of on pop?

[remyschwab]

Right now, a new amino acid or nucleotide is added to the kmer when a CollectionItem is popped from the queue. I think this is causing absorption to happen slightly later than necessary. Maybe it would be smarter to shift the kmer over before pushing? And extract the subhash then? Idk....