This must be fairly common, and fairly easy to do, just with numpy in some sensible way. It wouldn't have to be highly performant in the first instance, but it should be vectorisable with matrix ops.

• FRE is basically RMS error between corresponding points, see in code here.
• code in: scikit-surgerycore/sksurgerycore/algorithms/errors.py (for example)
• test in: scikit-surgerycore/tests/algorithms/test_errors.py

This one is easier than pivot calibration, do this one first.