A small patch for the rebasing of the main directory under src

[GalKepler]

Just a small fix - updated the location of the ._version.py written by setup_scm. Fixes #63

[arokem]

Test data now lives here: https://zenodo.org/record/6721110#.YrXPyRPMLKp

[arokem]

The tests on this branch are still pretty hopelessly broken. I tried to work towards fixing them in GalKepler#3, but there are still failures I don't understand. I would need @oesteban to explain to me how the logo implementation works before I can make progress. In particular, how does the model fit method know that only some of the data is provided as training data (e.g., on this line) and to account for that fact when selecting which volumes conform with b0 requirements that are set (for example) on this line?

[oesteban]

@arokem It seems we should work on the documentation actually - if you can't wrap your head around how it works then I'm pretty positive that literally any other outsider will fall into that trap. I'm opening #76 for this.

[oesteban]

Check out #84. In particular, there was a bug using the gradient table of the full dataset instead of that of the training subsample (i.e., N - 1).

[arokem]

Thanks! That helps! See questions on #84 for followup.