Bugfix: manhattan.R now indexes with BP, not SNP which should be an optional column

[samuelejones]

As the title suggests, the manhattan function does not require a SNP column, but when creating the index (to handle the cases where there are missing chromosomes), it uses the CHR and (possibly non-existent) SNP column. I have replaced the call to d$SNP with d$BP, which is a required column and produces the same result as using d$SNP (when it exists).

I'm honestly not sure how this has been broken for 7 years without a fix...