Duplicate fastq records.

[TDMedina]

Need to uniq-ify the list of donor reads after each sample is processed.

Duplicate reads can still exist if, for example, a read in a variant context has a mate mapping to a different chromosome in a second variant context. In this case, the mate will also be fetched as a read in that variant context, and the read pair will be added twice.

Because reads are currently only checked for uniqueness per variant context, these duplicates are not caught.

[TDMedina]

Duplicates are now checked and removed before writing fastq files.