Units for reduce_point_density

[dopplershift]

Given that we use units for everything else, it would be good if we could pass units properly to reduce_point_density. This is complicated by the fact that usually the station x,y locations are coming out of CartoPy projection functions, which won't have units attached. The pragmatic solution would be to assume meters for x,y unless otherwise specified.

[zbruick]

I just tried passing in units with the existing tests for this function and all of the existing tests passed. If you'd like me to work on this, let me know what the issue is that you're seeing.

[dopplershift]

Our workshop materials, and the MetPy examples, don't use units, they just have the constant of 300000. This is what the MetPy example puts out currently:

What I want is to pass in 300 * units.km for the radius and attach proper units to what comes out of CartoPy's transform_points() method, and hand that reduce_point_density. What it currently does:

That's no good. I'm guessing the SciPy kdtree code we're using is stripping units, so we need to massage units before passing in magnitudes. Since we're doing this using a class from SciPy and then a method call, I don't think we can use any wrapping machinery from pint.