DEPR: _metadata propagation

[jbrockmendel]

• closes #xxxx (Replace xxxx with the GitHub issue number)
• Tests added and passed if fixing a bug or adding a new feature
• All code checks passed.
• Added type annotations to new arguments/methods/functions.
• Added an entry in the latest doc/source/whatsnew/vX.X.X.rst file if fixing a bug or adding a new feature.

xref #51280

[jbrockmendel]

Mothballing, will revisit in a few months to see if the pro-_metadata folks make progress on getting it fully working.

[jorisvandenbossche]

Mothballing, will revisit in a few months to see if the pro-_metadata folks make progress on getting it fully working.

It is working ...

[jbrockmendel]

137 xfailed tests in tests.generic.test_finalize say otherwise.

[jorisvandenbossche]

The existence of a many packages implementing subclasses that happily make use of _metadata says otherwise?

The tests you are referring to use attrs to test this, and not all those tests are relevant for the subclasses using _metadata. For example, describe is xfailing but I would say this is fine to not propagate attrs/metadata. squeeze is xfailing, but that's only because it doesn't actually return a pandas object but a scalar. Some reductions are failing, but I think many subclasses don't care about that use case to preserve the class. Idem for categorical accessors.

So I would say that those xfailed tests are not really an indication of how well this is working in general.

[jbrockmendel]

There are better uses of both our time; I have PRs I've been asked to review, and I've asked you to look at topper's reduction PR.