Do not perform selection on non-genetic covariates

[Benjamin Chu]

Currently, sparsity is enforced on both the genetic and non-genetic covariates. It seems the standard is to not select for non-genetic covariates. So we need one of the following:

1. Change projection routine to exclude non-genetic covariates from being projected
2. Add routine to automatically regress out the effect of non-genetic covariates before running IHT

Probably option 1 is better

[Benjamin Chu]

resolved by #39