Add example for applying cross-validation with EQLHarmonic

[santisoler]

Description of the desired feature

Would be nice to have an example or a documentation page to show how to choose the best set of parameters while gridding gravmag data using EQLHarmonic. A good idea would be to start with one of the sample datasets and:

1. calculate gravity disturbance (using Boule),
2. project data into Cartesian coordinates,
3. use itertools to create set of parameters (damping and relative_depth),
4. use verde.BlockKFold or verde.BlockShuffleSplit to cross-validate the gridder with each set of parameters,
5. get the best predictor,
6. grid data onto a regular grid in geodetic coordinates (using the projection argument on the EQLHarmonic.grid() method),
7. mask distant values with vd.distance_mask.

Maybe we would need to block-reduce (or crop) the original data to speed up computations, specially if CIs will have to run the example.

Are you willing to help implement and maintain this feature? Yes

[santisoler]

For anyone wanting a tutorial for gridding gravmag, they can take a look at the notebook I presented at Transform2020: https://github.com/fatiando/transform2020-lightning/tree/master/santi

[mdtanker]

I was looking through the issues and noticed that this is still open, should it be closed with #305 and the following user guide https://www.fatiando.org/harmonica/dev/user_guide/equivalent_sources/eqs-parameters-estimation.html?

[santisoler]

Nice catch! Yes, we should close this now! Thanks @mdtanker!