Differential expression with single cell data

[arteymix]

Aggregating and generating processed vectors works well now.

The main difference with our subset analysis is that the subsets already exist prior to the analysis.

The analysis service must now derive the substring structure from the bioassay dimension. I've added some methods to infer which dimension a subset is part of, we just need to do the reverse: find which set of subsets applies to a given dimension. The QT can be used for disambiguation if necessary.

Another thing to be careful with is that upon deletion, the subsets should not be removed.