Add OSM powerplants as source

[brynpickering]

A spinout of powerplantmatching is OSM-powerplants which grabs powerplants identified in OSM.

[irm-codebase]

This would be good! Some things to keep in mind though:

This module currently does not use powerplantmatching or any similar tool for deduping. It is a "once source dataset" kind of deal for the time being, with most data coming from GEM.
I decided to keep it this way because ppm kept giving me issues, although some of them have been fixed already.
- PyPSA/powerplantmatching#233
- PyPSA/powerplantmatching#232

The current design "philosophy" is that it's better to miss data rather than duplicate it, at least until our ability to run histogram counterfactuals is strong enough to let users identify if mashing datasets together is doing something weird.
Dano Hawkins (@ddahawkins-TUDelft) already contributed algorithms to aid in this.

However, some additional comparisons and processing are still needed IMO.
We should, ideally, be able to compare the trend of the disaggregated data + imputed rooftop PV against national aggregates over for recent decades. They are bound to differ, but they should still show similar dynamics, at least...