RFC: Release GIL in vectorized SGP4 wrapper

[wrigby]

I was reading through the source to understand if the Python Global Interpreter Lock is released when calling sgp4 on a large SatrecArray. From what I can tell, the C extension currently holds onto the GIL, which blocks any other Python threads from doing work while the computations are happening.

I don't spend much time in Python C-extension land, so I'd love some direction on the best way to build and test this well, or even feedback that this is a bad or unnecessary idea.