Fix for xarray 2024.09.0

[aulemahal]

Well, this could have been expected from my decision to use a private function that I myself removed... But I had forgotten that we were explicitly using it.

So this fixes xclim with xarray 2024.09.0 by having a try-except for the import of the decimal year stuff.

Also, removed the deprecated function.

As of xarray 2024.09, the "decimal year" can be calculated through times.dt.decimal_year, no need to use a private function.

[Zeitsperre]

Sounds good, but we need to make a new release, unfortunately.

[aulemahal]

I'll let you decide how to manage the git stuff 😅

[Zeitsperre]

OK, we're going to need to rebase this PR onto the v0.52.1 tag, if that's alright. If you want to do the following, this would be helpful:

1. Start a new branch based on the v0.52.1 tag (v0.52.2).
2. git cherry-pick your commits to the new branch.
3. Open a PR and do not merge main into it

I can take things from there.

Scratch all of that. Working from here, see if you can get things working. There should be exactly two failing tests in all builds at the moment. I'll mark.xfail them here (they're removed in main) so that we can verify there aren't slower builds failing with unexpected failures.

[aulemahal]

There's a bug with intersphinx for statsmodel, maybe just a retry could solve it ?

And the notebook CI doesn't seem aware that it should use the old version of xclim-testdata...

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB