Robust support for data.frame inputs to multiGSEA()

[lianos]

A data.frame of features should be a fully featured input type for x in the multiGSEA() function.

We should be able to use data.frame inputs to allow the user to use any of the columns to ranke the features (with an appropriate ascending/descending parameter) for the variety of preranked GSEA methods.

The user should also be able to specify a logical column (perhaps the default one is "significant") to enable enrichment type of methods like goseq.

Currently we have a hack to support data.frame inputs by providing a named vector of scores as as the x parameter, and passing a data.frame of feature information via the xmeta. to decorate the features with more information (which happens in the calculateIndividualLogFC() function.

We'll need to:

• [ ] enable the multliGSEA,x parameter to be a data.frame of features with meta info.
• [ ] have the user specify which numeric column in x to use for ranking, along with the direction of sorting, so preranked GSEA methods can be seamlessly run (perhaps also some flag to indicate what to do with ties)?
• [ ] enable the user to specify a logical column to use for goseq-like enrichment analyses, where we need to define "features of interest" vs the background/universe (which is the features in the data.frame itself)