Refactor RMSE_test calculation using pygmt.grdtrack

[weiji14]

Simplifying and hopefully speeding up the code used to get interpolated grid values at certain groundtruth points. Previously used !gmt grdtrack which is a bash hack, but now there's pygmt.grdtrack that is a whole lot nicer! Instead of having temporary .csv and .nc files, we could simply pass in a pandas.DataFrame and xarray.DataArray in memory to pygmt.grdtrack, and get a pandas.DataFrame output we can analyze straightaway!

TODO:

• [x] Hotfix missing-one-row from pandas.csv load, oh the I/O misery! (1f1689820b1f5e2e207ae24eebab409a9f550e4a)
• [x] Input pandas.DataFrame directly into grdtrack instead of using intermediate .xyz file (e44c7c68b63fa56aa9388f83c81340a6c8981e70)
• [x] Attempt to input xarray.DataArray directly into grdtrack instead of using intermediate .nc file (66b97d2ad445cb70e833346df90549e12bd0f565)

[review-notebook-app[bot]]

Check out this pull request on ReviewNB: https://app.reviewnb.com/weiji14/deepbedmap/pull/149

Visit www.reviewnb.com to know how we simplify your Jupyter Notebook workflows.