In read.cross, use data.table::fread rather than read.csv

[kbroman]

data.table::fread (see the data.table package) is way faster than read.csv. We should use it in read.cross.