Create examples of dropping NaN rows then calling from_features

[PeterDSteinberg]

See @gbrener 's work here in PR #39 .

Make an example of a problem that

• needs to drop NaN values, such as the margins of a satellite image swath or the removing oceans from a raster related to terrestrial data, and
• show how from_features is convenient for reshaping back to the raster's approximate coords/dims. Then
• Visualizing the output of from_features that should show the NaN's in the right places

[gbrener]

Good starting point: https://github.com/ContinuumIO/xarray_filters/blob/master/xarray_filters/tests/test_reshape.py#L123-L158
I should be able address this soon - @PeterDSteinberg , do you by any chance have an example dataset handy?

[PeterDSteinberg]

@gbrener Maybe solve this issue at the same time as this Holoviews 2035 issue with soils data from NLDAS. Doing so would be helpful also for AGU prep

[PeterDSteinberg]

Hi @gbrener A faster way to get this testing done for NaN rows in reshaping is to just use this synthetic data function ts_clustering_example we have in the test_data.py of this repo (call it, then to_features, then add some NaNs for the test).