Duplication metrics parsing ignores multiple library preps

[b97pla]

In case duplicates have been marked on a bam file with pooled results from multiple library preps (which is the case for >60X samples), mark duplicates reports duplication rates for each prep separately but only the first prep is parsed in https://github.com/NationalGenomicsInfrastructure/ngi_pipeline/blob/master/ngi_pipeline/engines/piper_ngi/parsers.py#L159.

Example:

## METRICS CLASS picard.sam.DuplicationMetrics
LIBRARY UNPAIRED_READS_EXAMINED READ_PAIRS_EXAMINED UNMAPPED_READS UNPAIRED_READ_DUPLICATES READ_PAIR_DUPLICATES READ_PAIR_OPTICAL_DUPLICATES PERCENT_DUPLICATION ESTIMATED_LIBRARY_SIZE
SX612_UL-80-93-Buccal-seqlib.v2 1439015 430201701 42575769 564421 69043442 12580258 0.160878 1401844231
SX612_UL-80-93-Buccal-seqlib.v1 1695386 340337028 35523482 629611 55482498 13173994 0.16354 1153360123