Thank you for your contribution to Astropy! 🌌 This checklist is meant to remind the package maintainers who will review this pull request of some common things to look for.

• Do the proposed changes actually accomplish desired goals?
• Do the proposed changes follow the Astropy coding guidelines?
• Are tests added/updated as required? If so, do they follow the Astropy testing guidelines?
• Are docs added/updated as required? If so, do they follow the Astropy documentation guidelines?
• Is rebase and/or squash necessary? If so, please provide the author with appropriate instructions. Also see instructions for rebase and squash.
• Did the CI pass? If no, are the failures related? If you need to run daily and weekly cron jobs as part of the PR, please apply the "Extra CI" label. Codestyle issues can be fixed by the bot.
• Is a change log needed? If yes, did the change log check pass? If no, add the "no-changelog-entry-needed" label. If this is a manual backport, use the "skip-changelog-checks" label unless special changelog handling is necessary.
• Is this a big PR that makes a "What's new?" entry worthwhile and if so, is (1) a "what's new" entry included in this PR and (2) the "whatsnew-needed" label applied?
• At the time of adding the milestone, if the milestone set requires a backport to release branch(es), apply the appropriate "backport-X.Y.x" label(s) before merge.

While the code looks good, locally I see a degraded performance, which I didn't expect. I'll need to run some more tests before merging.

Timings for two arrays of shape (2000, 2000):
main: 243 ms ± 533 μs
pr: 357 ms ± 947 μs

Timings for two arrays of shape (4000, 4000):
main: 900 ms ± 1.85 ms
pr: 1.33 s ± 2.56 ms

The old code here does not look like it is subject to the future "shape" deprecations. It already uses reshape and shape is only used to retrieve the shape of the inputs.
If this is correct and given the differences in the timings I suggest we don't merge this PR. The reason is we constantly strive to improve performance of GWCS and this will degrade performanec of all imaging transformations.

@nden - wow, I would have thought my PR would have sped up, not slown down. I'd like to investigate, since perhaps this means that on the numpy side this could be improved a lot too. Just so I'm sure that I'm doing the right thing, could you give the full example you tested? In the meantime, I'll convert this to draft so it won't get merged accidentally.

The old code here does not look like it is subject to the future "shape" deprecations

That is because it was already adjusted in #18562 - when reviewing that PR, I thought I could make a further improvement...

FWIW

SOME BENCHMARKS HAVE CHANGED SIGNIFICANTLY.
PERFORMANCE DECREASED.

@mhvk Thanks for following up! The test example is

from astropy.modeling import rotations
import numpy as np

x, y = np.mgrid[:4038, :4051]
n2c = rotations.RotateNative2Celestial(5.6, -72, 180)

timeit n2c(x, y)

Great, thanks both!