plink genotype conversion to BED not necessary

[lcpilling]

Package used to use plink v1.9 which does not support BGEN format nicely. Therefore variant extraction from BGEN follows this process:

1. Subset Bulk BGEN using bgenix
2. plink conversation of BGEN to BED
3. plink conversion of BED to RAW (text file)

This has two issues:

1. Genotype doses (from imputed data) become hard calls
2. Multiallelic SNPs are dropped (only the first is kept if there are multiple variants at the same position)

Now that the package uses plink2 the BED conversion can be skipped entirely as plink2 can handle BGEN format

This fix is in the pipeline... once access to UK Biobank is restored.