Possible univariate feature selection/IHT memory issue

[egillax]

@FleurVereijkenMi is running out of memory I think (R session crashes) when running IHT on big data. I think (need to double check) that IHT should not be taking more memory than regular Cyclops.

[egillax]

This happens with 900k population and 3000 outcomes. Which should be fine.

[egillax]

not iht but univariate feature selection.

[egillax]

Happens both in IHT and UFS