Back to basics for diffuse fraction functions

[adriesse]

Is your feature request related to a problem? Please describe.
Our collection of diffuse fraction model functions (identified as DNI estimation models) is growing. Currently none of the functions supports the most basic usage I expect, which is to calculate diffuse fraction from kt.

Describe the solution you'd like
A set of diffuse fraction functions that take kt as input (and more kwargs if the model needs it) and produces diffuse fraction as output.
A higher level function could be provided to do the higher level task of decomposing ghi using a kwarg to select the model.

Describe alternatives you've considered
Implement my own versions of the functions locally, which I have done.

Additional context

#1128
#1179
#1685
#1705
#1730

[cwhanse]

@adriesse do you think we factor the diffuse fraction = f(kt) functions out of the various decomposition functions?

[adriesse]

@cwhanse that's the gist of it.