DSL2: Metagenomics: Allow Skip-pairmerging before running metagenomics

[merszym]

Problem

In dsl2/eager3 we currently dont allow the combination --run_metagenomics --preprocessing_skippairmerging because we are not handling paired reads in the metagenomics subworkflow

ToDo

One needs

1. Overhaul the bamfiltering module, so that paired reads are preserved (and tagged as such)
2. Then account for paired reads in the metagenomics subworkflows

[ilight1542]

now working on this on branch metagenomics-pairedend

current status: input into krakenuniq working using both a mixture of PE and SE job submissions, that then are all kept in the output directory ! :)

-- but PE data seems to cause the krakenuniq call to expect fasta files not fastq files. haven't figured that out.

still todo: maltextract