Track unmatched variants in match_variants

[smlmbrt]

Currently we don't have indications of why certain scoring file variants didn't match. We should track/output whether they did match but were excluded because they were:

• Strand ambiguous (A/T, C/G)
• Matched to a multi-allelic variant that was excluded
• Strand flip (that we didn't correct because of the flag)
• etc.

[smlmbrt]

Only partially addressed by #13, still outstanding:

• Allow multi-allelic matches to occur instead of removing them from the target. Use a label function to select whether these matches are excluded from the final scoring file. [The labelling of duplicate best_match function we already implemented should handle this case]
• Allow flipped matches for scoring file lines without better matches. Use a label function to select whether these matches are excluded from the final scoring file.
  • Add information about whether the match has been flipped to the [dataset]_summary.csv file

[smlmbrt]

Closed in #26