different df col/row data types

[sebastian-mutz]

Currently, read_csv assumes real for all non label columns/rows. Instead, make col/row specific data types a df attribute, insert case before conversion (s2r) and convert as needed.