I’ve investigated something in this regard in the past few weeks but with a focus on *MASS and 1D beams. The results are in this other thread. Here I verified that no matter the eccentricity of the mass, the frequency results remain unchanged, so no rotary inertia is considered in CalculiX formulation. Maybe some valid workaround is to connect a very stiff beam (E=99999GPa for example) between the mass element and the structure you are interested in. However, I did not try this yet.
Do the frequency results change when you modify the distance of the point mass? When I tried this with 1D beams, nothing changed. I am curious about what happens if one uses solid elements, my intuition is that if it did not work for expanded solids, it would not work for intrinsic ones…