Packaging fixes

[Zeitsperre]

Pull Request Checklist:

• This PR addresses an already opened issue (for bug fixes / features)
  • This PR fixes Reduce the size of /docs #247
• (If applicable) Documentation has been added / updated (for bug fixes / features).
• (If applicable) Tests have been added.
• This PR does not seem to break the templates.
• HISTORY.rst has been updated (with summary of main changes).
  • Link to issue (:issue:number) and pull request (:pull:number) has been added.

What kind of change does this PR introduce?

• Makes it so that tests are not treated as an installable package.
• Removes the tests_require in the setup.py.
• Adds a new dev recipe for convenience.
• Includes the documentation in the source distribution, excludes the NetCDF files.
• Renames the pytest marker from requires_docs to requires_netcdf.
• Substantially reduced the size of the files in /docs/notebooks/samples (from 16Mb to 400kb) and changed the Notebooks and tests to work around it.
• datablock_3d moved to a new xscen.testing
• xscen.testing.fake_data to easily populate DataArrays with fake, but realistic data.

Does this PR introduce a breaking change?

Sort of. tests no longer acts like an installable module. This wil only affect xscen developers who write tests.

Other information:

This will be fully obsolete when we move to pyproject.toml, but, for those curious, why not use tests_require?

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[RondeauG]

@aulemahal @juliettelavoie No need to review everything, but I'd like you to look into the Notebooks, at least. I had to change them somewhat because we reduced the size of the underlying datasets.

• I created a new fake_data function to fake 70/120 years of climate data after extract_dataset, for Getting Started and Warming Levels. That means that we can do things properly, but keep the on-disk files at a very small size.
• pangeo-cmip6.csv is now the bare minimum (48 lines vs 35000!) to make the Catalogs notebook work.
• The Ensemble Reduction notebook now uses precomputed tg_mean from xclim.testing

[RondeauG]

LGTM, but I'll wait for the others to look at it too (for my changes, at least).

[aulemahal]

Looks good to me.