Support for double precision float

[TheCount]

This request is similar to #23. Some devices store values as 64-bit floating point in four consecutive registers, for example the Siemens PAC2200 energy meters store their demand/supply Wh values as double precision floats, so some -t 3:double type would be nice.