Investigate issues with move to numpy 2

[mferrera]

We have observed issues since dropping numpy 2 support in #1285

With numpy 2 + pandas 2.0.2 (the RMS 14+ environment) we see quite a few

ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

With numpy 2 + pandas 2.2.1 (latest), we see much less of them, but they become:

ValueError: could not broadcast input array from shape (87,) into shape (86,)

These issues do not occur with numpy<2.

So far this has only popped up in fmu-tools: https://github.com/equinor/fmu-tools/actions/runs/12681226680/job/35344528453

[mferrera]

You need pandas 2.2.2 to use numpy 2.0.

This may indicate we need to maintain a separate branch for RMS environments, cf https://pandas.pydata.org/pandas-docs/stable/whatsnew/v2.2.2.html

[mferrera]

For fmu-tools solution is just to separate its pandas 2.0.2 mandated test to a separate job where the correct numpy version is installed. We can probably close this as a "warn" if using numpy 2 + pandas < 2.2.x, which aren't compatible (i.e. it's a pandas problem)