Consider OffTargetDetector._to_amplicons specifying alignment strand for primer pairs

[ameynert]

If the primer pair is oriented with the left primer on the positive strand and right primer negative, the amplicon span should have Strand.POSITIVE.

GGTCCAGTTCAAGTGCTGGGAGAGCATCCTCCACAAGGTCTAGTGGTATGGTGGT
L>>>>>>>>>>                                                                                    <<<<<<<<<<R

If the primer pair is oriented with the left primer on the negative strand and the right primer positive, the amplicon span should have Strand.NEGATIVE.

ACCACCATACCACTAGACCTTGTGGAGGATGCTCTCCCAGCACTTGAACTGGACC
R>>>>>>>>>>                                                                                    <<<<<<<<<<L

(randomly generated sequence used as example).

[clintval]

I would support this

From what I can tell, when generating an amplicon span from a PrimerPair, we inherit the strand of the primers during replacement of the end property of the left primer span:

prymer/prymer/api/primer_pair.py

Line 252 in 5d79a52

return replace(self.left_primer.span, end=self.right_primer.span.end)

But we don't account for strand when generating an amplicon span from a pair hit:

prymer/prymer/offtarget/offtarget_detector.py

Line 425 in 5d79a52

amplicons.append(Span(refname=plus.refname, start=plus.start, end=minus.end))