The effect of n_pvrows on edge effects for pvfactors_timeseries should be documented

[kandersolar]

The pvfactors model treats rows as infinite in length but finite in number, with the current default being n_pvrows=3. With such a small number of rows come interesting edge effects such as fully-illuminated ground underneath the array at low sun elevation. This particular effect creates morning and afternoon spikes in rearside irradiance that shrink as n_pvrows is increased:

Increased row count, and therefore increased ground shading, also reduces rearside irradiance at midday when the rear surface is exposed to the entire ground plane.

Whether the default value of n_pvrows should be changed so the output is more representative of the interior of large arrays is debatable, but regardless of what n_pvrows defaults to I think we should somehow document its effect on output. Maybe just a sentence or warning in the function docstring? Maybe a gallery example with a plot like the one above?

Probably worth pointing out that increasing n_pvrows significantly increases runtime. See SunPower/pvfactors#134.