This step concatenates files and then feeds them into filter_reads which also needs all the unaggregated files and the aggregates are never used again. We can just do the aggregation at the start of filter_reads or more likely not at all (iterate over input read id files)