Fix truncated reads

[lskatz]

• Took advice from @BioWilko to fix truncated reads by not using the --gzip option on fastq-dump in #13
• Hashsums changed as a result for all datasets
• Hashsums varied with gzip metadata and so we had to check hashsums on uncompressed reads
• Hashsums have been recalculated accordingly
• Unit testing filled up the available space for one dataset and so unit testing has been altered to chunk 20 at a time