Couple of numpy-related fixes

[davidwilby]

📝 Description

We have a few tests routinely failing at the moment for a handful of reasons mostly relating to numpy 2.

One such error is due to DeepSensor's dependency stheno not yet pinning the version of fdm which repairs some of the numpy compatibility bugs. I've raised a PR in that project here: wesselb/stheno#27

In this PR there are a couple of corrections to alleviate test failures:

1. Use of np.product in some tests, this was changed in something like numpy 1.16 to np.prod.
2. The type of encoder_scales was being set to a numpy float32 with numpy 2 (possible relating to https://numpy.org/devdocs/numpy_2_0_migration_guide.html#changes-to-numpy-data-type-promotion) which was causing the saving and loading test to fail. I've added an int() but unsure if this is appropriate/ideal, though all of the tests pass locally.

✅ Checklist before requesting a review

(See the contributing guide for more details on these steps.)

• I have installed developer dependencies with pip install -r requirements/requirements.dev.txt and running pre-commit install (or alternatively, manually running ruff format before commiting)

If changing or adding source code:

• tests are included and are passing (run pytest).
• documentation is included or updated as relevant, including docstrings.

[tom-andersson]

Awesome, thanks so much for catching these niche deprecation bugs @davidwilby!

You mentioned changing encoder_scales to an int but I can't see that in this PR. Looks like the tests are passing here, so maybe it's not an issue? I see that numpy 2.2 is being used by the test worker.

encoder_scales has to be a float though, not an int, so we'd have to find another solution if this is an issue.

[tom-andersson]

@all-contributors please add @davidwilby for bug

[allcontributors]

@tom-andersson

I've put up a pull request to add @davidwilby! 🎉

[davidwilby]

You mentioned changing encoder_scales to an int but I can't see that in this PR. Looks like the tests are passing here, so maybe it's not an issue? I see that numpy 2.2 is being used by the test worker.

Sorry, I mistyped, I meant float and explained myself poorly! model.defaults.gen_encoder_scales calls data.utils.compute_xarray_data_resolution where this fix is made.