Allow "per-sample" processing

[jpfeuffer]

Useful for huge studies where the samples are actually unrelated. Unrelatedness could be one or more of the following:

• Different number of fractions
• Different instruments
• Unrelated conditions where you do not want to compare across

I was thinking about adding a param like "splitting-column" to define a column in the sdrf or experimental design where we split by value in the very beginning and then process them in a loop through the whole workflow.

This needs quite some restructuring though since we probably need to pass through the "value" to which each file belongs and in the proteomicslfq step collect the ones that belong to the same "value"

[jpfeuffer]

@timosachsenberg thoughts?

[jpfeuffer]

See also #27