Add argument to `drop_uninformative_columns()` to provide R code to only select columns identified as uninformative.

[joethorley]

Can then use as drop in that robust to changes in the underlying data