fix the names of the floating-point operations in the Cmm parser