More streamlined unit conversion

[barneydobson]

From discussion at 31

This might be debatable, but I think unit conversion functions should go on their own and be functions (in their own module), not constant conversion factors:

def m3perh_to_m3pers(value: float) -> float:
    factor = 1 / 60 / 60
    return value * factor

q = m3perh_to_m3pers(tot * pipe_design.precipitation)

This is more readable, easier to re-use and to test.

I have to deal with units in every project, it's so annoying. I found something called quantities. The use case above would be something like:

import quantities
def convert_unit(value: float, from_unit: str, to_unit: str):
   from_unit = quantities.CompoundUnit(from_unit)
   to_unit = quantities.CompoundUnit(to_unit)
   return (value * from_unit).rescale(to_unit).magnitude

# Example
>>> print(convert_unit(1, "m**3/h", "m**3/s"))
0.0002777777777777778 
>>> print(convert_unit([1,2], "m**3/h", "m**3/s"))
[0.00027778 0.00055556]

It's another dependancy, but it's quite a pleasing solution - what do you think? Maybe also cheginit would be interested.

Interestingly it looks like if you had any number implemented as quantities, it would just handle it automatically throughout.

>>> 1 * pq.CompoundUnit("m**3/h") + 0.003 * pq.CompoundUnit("m**3/s") 
array(11.8) * (m**3/h)

I agree that using a functional approach might be better. I think adding a new dependency depends on the frequency of unit conversion in the code. If you think, there will more of such instances, I recommend using something more mature like Pint that supports pandas and xarray.