Add support for float data type for 32-bit real floating point values. Treat double as a synonym for the 16-bit data type.