I would like to suggest applying the attached patch to ReflectionSymmetry to fix (I believe) its handling of the tensorparity attribute.
In ReflectionSymmetry/interpolate.c parities[dir] is set to the behaviour under reflection across the "dir"=0 plane of an object of type tensortypealis. It does not take tensorparity into account. Therefore check_dir[dir] must take tensorparity into account explicitly.
The second change is "parity = 1" instead of "parity = tensorparity" which states that if a point is never reflected then the value that was on the grid is to be used as-is.
I have tested these changes with the magnetic field created by a wire along the x,y,z axis and reflection across the various planes and with RotatingSymmetry180 and *90. Please see ticket
for sample parameter files and test scripts.