divide input sequences for positive / negative strand

[klamkiew]

based on ORF density, sequences can be (roughly, heuristically) grouped into sense and antisense strand. Sometimes, both strands are interesting, thus it should be provided.