BigFloat add_mul and sub_mul using MPFR

[nsajko]

After #211 it might also make sense to implement the three-argument BigFloat versions of the add_mul and sub_mul operations in terms of mpfr_fma and mpfr_fms.

[blegat]

Indeed, I didn't know that there were these in MPFR, it would definitely make sense