big integers are parsed as Float64

[mhinsch]

I have a tsv file that contains object ids (as produced by objectid) as the first two columns. When reading with CSV.read(fname, DataFrame), the first column is correctly recognised as Int128 while the second column is read as Float46.