Utility function to create kdtrees

[leouieda]

We support using pykdtree as a replacement for most kdtree operations
but the feature set is the not identical to the scipy implementation.
We had no way of dealing with this before and had some repeated
try...except code for importing the modules.

A new kdtree function in verde/utils.py centralizes the imports,
allows using the scipy kdtree even if pykdtree is installed, and
handles transposing the coordinate arrays before creating the tree.

Reminders

• Run make format and make check to make sure the code follows the style guide.
• Add tests for new features or tests that would have caught the bug that you're fixing.
• Add new public functions/methods/classes to doc/api/index.rst and verde/__init__.py.
• Write detailed docstrings for all functions/classes/methods.
• If adding new functionality, add an example to the docstring, gallery, and/or tutorials.

[djhoese]

Just curious, what features in the scipy kdtree are not supported in the pykdtree? Are there bugs in pykdtree that are making these things not work? Or did scipy add new features in the last year or two?

Edit: Also pykdtree now has a mask keyword argument you can provide to the query call to mark certain neighbors as bad. This lets you reuse the same KDTree even if your neighbor masks change.

[leouieda]

@djhoese no bugs, just the other class methods besides query: https://docs.scipy.org/doc/scipy/reference/generated/scipy.spatial.cKDTree.html

I don't need them right now but I will need query_ball_point soon.