Orix 0.13.1 Causes Test Failure in Pyxem

[CSSFrancis]

Describe the bug
It seems like orix 0.13.1 causes a test failure in the orientation mapping code. (see the action: https://github.com/pyxem/pyxem/actions/runs/11052306498/job/30704105695)

It might be just a trivial example of equivilent orientations or some small change there. I haven't looked at it in too much detail. @hakonanes if you have an idea off the top of your head that might cause that change that might be helpful otherwise I can look into it.

[hakonanes]

If you can wait until Saturday I can have a look. Seems like some inconsistency in the calculation of misorientation angle from 0.13.0 to 0.13.1.

[hakonanes]

The test is too complicated for me to debug. I don't know what the variables orientations, rotations, or orients express. I'd appreciate it if you could give a minimal working example.

[CSSFrancis]

@hakonanes no worries, I'll look into on Monday and I can track down the change there :)

I was just thinking you might have a quick idea for what changed

[hakonanes]

Unfortunately not. We didn't have to update any misorientation angle tests when making numpy-quaternion an optional dependency. So I didn't expect this to be a problem...

[hakonanes]

From highly related #539:

@ericpre:

In hyperspy/hyperspy-extensions-list#56, I couldn't get the test suites of pyxem and kikuchipy to run without segmentation fault when orix (and dependencies) is installed using packages from pypi.org

Installing orix from conda-forge work as a workaround.

Me:

Thanks. Will try to dig out the offending Numba functions from the test logs.