Add non-weighted pressure-based mean function

[dopplershift]

While we have mean_pressure_weighted, we should add a corresponding integral-based mean calculation to go along with it. For instance, something like bunkers_storm_motion should probably be using that instead of numpy.mean, since the latter would be biased by uneven sampling of levels in a vertical profile. It should be as simple as taking mean_pressure_weighted and ripping out the weighting (and adjusting the denominator to p[0] - p[-1]).

Not sure what to call it...mean_vertical? vertical_mean? mean_non_weighted?