Changes to PVGIS in version 5.2

Describe the bug PVGIS is releasing version 5.2 on March 1st, 2022. In the new version, the irradiance databases CMSAF and COSMO have been removed. Additionally, the ERA5 coverage has been extended to cover almost the entire globe and a new database, SARAH-2, has been added.

The full description can be found here.

The new default databases can be seen here:

Additional context There is nothing that indicates any changes to the API, so the pvlib.iootools functions get_pvgis_tmy and get_pvgis_hourly should continue to work.

However, some of the test files used in test_pvgis.py are from the CMSAF database. Since those tests seem to be either based on local files or mock tests, I do not expect test failures. However, it would probably make sense to update the tests regardless.

[kanderso-nrel]

Probably a good idea to scrub mentions of the retired databases from the docstring as well:

https://github.com/pvlib/pvlib-python/blob/c2431832040f585969bbd1868f6acdd6ddb4e0a5/pvlib/iotools/pvgis.py#L140-L144

@AdamRJensen thoughts on tagging this issue for 0.9.1?

[cwhanse]

@AdamRJensen do you know if the content of CMSAF and COSMO is migrating to the new database? Sorry that I’m not familiar with PVGIS.

I missed the link to the documentation in my quick read of your issue, the documentation answers my question. I agree it would make sense to replace the test files or (more simply) to add a note to the test code identifying the data source.