ADR Suggestion - Arithmetic operations for Parameters and DescriptorNumbers

[damskii9992]

General

Adding capability to do arithmetic operations on Parameters and DescriptorNumbers allow users to use fitted parameters to do quick and easy calculations with uncertainties and units to calculate items of interest, as well as quickly and easily create new parameters/descriptors from previous ones.

Current implementation

This is not currently implemented

Proposed implementation

Add the following magic methods to Parameter and DescriptorNumber:

__add__
__sub__
__mul__
__truediv__
__pow__
__neg__
__abs__

The output should be a Parameter with automatically calculated min/max if any of the elements of the arithmetic operation is a Parameter, otherwise the output should be a DescriptorNumber. In either case the unit and variance is also automatically propagated.

The operations should work between Parameter and DescriptorNumber as well as with floats and integers.
For addition/subtraction with scalars, the variance is unchanged, unless the scalar is a DescriptorNumber with a variance, in which case the variances are added, and the min/max of Parameters are shifted accordingly.
For multiplication/division with scalars, the variance is also multiplied/divided together with the min/max.

Implementation details

To allow the arithmetic operations between Parameter and DescriptorNumber, as well as with scalars, in both directions, i.e. a - b = c and b - a = d, the reverse magic methods also needs to be implemented:

__radd__
__rsub__
__rmul__
__rtruediv__
__rpow__

These methods are called on the 2nd operand when the 1st operand does not know how to do the operation (which the interpreter knows when NotImplemented is returned), or when the 2nd operand is a subclass of the 1st operand, such as in the case of DescriptorNumber and Parameter.

When the unit of the elements of the addition/subtraction operations differ, yet have the same physical dimension, the 2nd element is converted to the unit of the 1st element before the operation, (i.e. 2m + 2cm = 2.02m and 2cm + 2m = 202cm).

This behavior differs from pure Scipp behavior, which does not allow addition/subtraction between different units of the same dimension, but mimics behavior of other unit libraries such as pint. This point is also discussed in Scipp: scipp/scipp#3456

For multiplication/division, the derived units from Scipp with numerical prefactors: (e.g. 2m * 1dm = 2 0.1m^2) is automatically converted to their base unit without the prefactor, i.e. 2m * 1dm = 0.2 m^2. There is also an open dicussion on this issue on Scipp: scipp/scipp#3500