Add support for floating point types

[dimitribouniol]

This would include big and little endian representations of Float16, Float32, Float64, and Float80.