DOC: irradiance.dni() not exposed in API, but also not private

[mikofski]

Describe the bug
there is a nice function in the code called pvlib.irradiance.dni() that makes sure (GHI-DHI)/cos(zenith) isn't bonkers when cos(zenith)->zero, but it's not in the documentation.

Ok it's in one location, here

I think it just needs to be listed in the api.rst page, but where? It's not really a DNI estimation method, because it uses exact values of GHI and DHI, so not sure where it should go

[mikofski]

@patrickjmathiesen this might be a good first PR for you if you're interested?