cf-units wheel on pypi not compatible with numpy2

[heikoklein]

🐛 Bug Report

Installing numpy-2 together with cf-units from pip is possible, but when running it, the following error is given:

venv/lib/python3.11/site-packages/iris/common/mixin.py:10: in <module>
    import cf_units
venv/lib/python3.11/site-packages/cf_units/__init__.py:23: in <module>
    from cf_units import _udunits2 as _ud
cf_units/_udunits2.pyx:1: in init cf_units._udunits2
    ???
E   ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

How to Reproduce

Steps to reproduce the behaviour:

1. pip install "numpy>=2"
2. pip install cf-units
3. pip list

   Package                Version
   ---------------------- -------
   antlr4-python3-runtime 4.7.2
   cf-units               3.2.0
   cftime                 1.6.4
   Jinja2                 3.1.4
   MarkupSafe             2.1.5
   numpy                  2.1.1
   pip                    22.0.2
   setuptools             59.6.0

4. python -mcf_units

   Traceback (most recent call last):
   File "/usr/lib/python3.10/runpy.py", line 187, in _run_module_as_main
   mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
   File "/usr/lib/python3.10/runpy.py", line 146, in _get_module_details
   return _get_module_details(pkg_main_name, error)
   File "/usr/lib/python3.10/runpy.py", line 110, in _get_module_details
   __import__(pkg_name)
   File "/home/heikok/cfunits2/lib/python3.10/site-packages/cf_units/__init__.py", line 23, in <module>
   from cf_units import _udunits2 as _ud
   File "cf_units/_udunits2.pyx", line 1, in init cf_units._udunits2
   ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

   Expected Behaviour

   cf-units should work together with numpy2. or cf-units should be marked incompatible with numpy2

Environment

• OS & Version: Ubuntu 22.04 LTS
• cf-units Version: 3.2.0

[ocefpaf]

Needs https://github.com/SciTools/cf-units/pull/427 In a way, past builds will require a numpy<2 pin to keep working.