Restrict shapely dependency to <=2

[kandersolar]

Closes #126.

I wonder if we should think about releasing 1.5.2 soon? Besides these deprecation warnings and the tilt=0 issue from #125, it would be nice for the latest pvfactors version to allow pvlib==0.9.0 on installation. Someone asked about this on the pvlib google group: https://groups.google.com/g/pvlib-python/c/hNaNsM1pNSw

[kandersolar]

Hi @campanelli-sunpower, just a bump, especially about cutting a new release that doesn't forbid the latest pvlib :)

[campanelli-sunpower]

Because pvfactors is a library, I think we should allow upgrading of Shapely until (but not including) v2, at which point the warnings will become failures.

[campanelli-sunpower]

Thanks @kanderso-nrel .

[kandersolar]

I suspect the test failures are caused by the numpy change discussed in pvlib/pvlib-python#1369

If that's true (need to verify it locally), I'd advocate for the same fix: weaken the test tolerance a bit to allow for the variation from numpy

[campanelli-sunpower]

@kanderso-nrel Looks like some small numerical drift on the newer Pythons. I think it would be ok to widen the tolerance slightly.

Note the recommendation here to not use numpy.testing.assert_almost_equal:

It is recommended to use one of assert_allclose, assert_array_almost_equal_nulp or assert_array_max_ulp instead of this function for more consistent floating point comparisons.

[kandersolar]

@campanelli-sunpower Shall I update this PR with those changes?

[campanelli-sunpower]

@kanderso-nrel Please do so if you are ok with that. It would keep the review sequence more straightforward. Thanks.

[campanelli-sunpower]

Thanks for updating these test methods too.