Catch a (0,) dimensionality issue on spectral axis in beamcon_3D

[tjgalvin]

I recently installed a new environment for some racs-all testing and started to run into a strange issue.

File "/datasets/work/jones-storage/work/miniconda/miniforge3/envs/flint_racsall/lib/python3.12/site-packages/racs_tools/beamcon_3D.py", line 768, in initfiles
    crpix = int(spec_axis.wcs.crpix)
            ^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: only 0-dimensional arrays can be converted to Python scalars

Digging into this is appears as though a numpy array of length 1 is being returned in the spec_axis.wcs.crpix WCS instance. The attempt to int therefore fails, even though the array was of length 1. I am not sure if this was a change in numpy or astropy, as it honestly could be either. I am relatively certain it is not a fitscube issue as the cube there has gone through the astropy.io.fits infrastructure, and read in as such correctly as well.

This PR has a type check, and if a numpy array is found it is first squeezed before passing through the int typecast. I believe it is a backwards compatible change, i.e. whatever changed in whatever dependency will contrinue to work.

[AlecThomson]

All looks pretty reasonable to me! I'm going to make a slight change from a type check to a scalar check. Any idea which version of astropy+numpy caused this issue?