Update xarray and friends

[scottyhq]

There a many great new xarray features from the last year. This bumps to the latest version as well as optional dependencies

[scottyhq]

/condalock

[github-actions]

👈 Test this PR on Binder

[weiji14]

Do you know whether the migration of datatree into xarray is good enough yet to just drop xarray-datatree? I saw in the changelog at https://github.com/pydata/xarray/releases/tag/v2024.05.0 which said that most of the code has been migrated, but not sure how feature complete things are.

[scottyhq]

true! time to drop it here

[scottyhq]

/condalock

[scottyhq]

Could not solve for environment specs
The following packages are incompatible
├─ scipy 1.9.3  is requested and can be installed;
└─ xarray >=2024.05.0  is not installable because it requires
   └─ scipy >=1.10 , which conflicts with any installable versions previously reported.

Situations like these make me question the pinning strategy here, wouldn't it be better to unpin everything and just use the lockfile to keep track of exact versions over time?

[weiji14]

Yeah, we were slowly trying to move towards >= or ~= pinning per #46, but could revisit the topic. I do think that conda-lock has matured quite a bit now, and we could consider just unpinning everything and rely on the unified lockfile (once I get the time to update #14).

[scottyhq]

/condalock