DEP: interpolate: deprecate interp2d

[ev-br]

Reference issue

mailing list discussion: https://mail.python.org/archives/list/scipy-dev@python.org/thread/LU72CAGYFH37PZWCANMUF6UWFNSPZRHU/

What does this implement/fix?

Make interp2d raise DeprecationWarnings starting from scipy 1.10

Additional information

Posting early to increase exposure: the mailing list thread needs at least two weeks (i.e., this can go in not earlier than Oct 20).
Also the transition guide link is for now to a github gist, will likely need to be updated based on the final state of gh-17178.

[ev-br]

@j-bowhay Jake, would you mind pointing me to how to keep track of the deprecation? Especially given the three-step process, as suggested by Matthew in the mailing list?

[j-bowhay]

The status of all deprecation if currently tracked in #15765 cc. @h-vetinari

[j-bowhay]

A few comments to hopefully make this more consistent with other deprecations. Also I think it would be nice to have a test checking that the warning appears when it is expected to.

[j-bowhay]

Hmm documentation not rendering quite right. Not sure if it's the new lines in the message or an issue with the indentation that's causing the problem.

[ev-br]

OK, I think it's easiest to just copy-paste the message into the docstring. Will do together with eating-own-dogfood exercise, there is interp2d usage in scipy.stats to be ported over.

[ev-br]

Test failures are due to internal interp2d usage in scipy.stats, which gh-17198 removes.

[ev-br]

Rebased on main now that gh-17198 is in, and lumped together review suggestions from Jake.

Apologies for the mass ping of code owners! This was a git mistake when pulling in changes made in the GH interface.

Recording the error here: apparently, when there is a review change via GH interface "suggested edit" and a local change, updating a PR needs two rebases: one origin/<PR fork> to pull in the reviewer edits, and the second on upstream/main to sync the PR branch with the upstream main.

[j-bowhay]

Other than the lint errors this looks good to go now

[ev-br]

No objections surfaced in the ML discussion, so I'd like to merge this for 1.10 before branching. Am going to hit the merge button this week unless further comments.

One remaining glitch is that the porting recommendations are still in a gist under my GH account. This can be adjusted in a follow-up though, if we figure a better place for it.

[ev-br]

OK, the mailing list discussion outcome is positive, and there are no further comments either in the ML or here. Let's press on with this deprecation. Thanks Jake for the help, thanks to everybody who contributed to the ML discussion. Merging.