Make the MultiQC report depend on the cutdapts report, not the trimmed reads

[arteymix]

This way, we could remove the trimmed reads without having to rerun the whole pipeline if we need to regenerate QC reports.

[arteymix]

One way of achieving this is to create a dummy task whose output is the cutadapt report of a particular TrimSample task and use that as a dependency for generating the report.

For this to work cleanly, we could add a structured output to the Cutadapt task which allows one to peek the report by attribute name as well as the trimmed read files.