Add option to output closest relative(s) per sample

[amkram]

Adds extract --closest-relatives which outputs a TSV line for each sample (selected with other options) with the following fields:

1. Target sample
2. Comma-separated list of the closest relatives (multiple entries if tied for closest)
3. Distance in branch length to target sample

If --break-ties is specified, only the lexicographically smallest sample ID is chosen in field 2 (recommended if selecting many samples to save space)

E.g.
matUtils extract --closest-relatives output.tsv [--break-ties] -i input.pb.gz