Bistatic RCS generated by dielectric cubes

Hi, I was reproducing the tutorial for dielectric spheres with bempp-cl v0.3.2 but this time I applied it to cubes:

sphere0 = bempp.api.shapes.cube(length = 0.4, origin=(-1.0, 0, 0), h=0.2)
sphere1 = bempp.api.shapes.cube(length = 0.4, origin=(1.0, 0, 0), h=0.2)

The interior solution (Squared total field strength on the right) is showing some unusual behavior—it seems like the fields are not correct:

Any suggestions? Could this be a bug, or is there something I might be misunderstanding?
I have attached the ipynb here.

Thank you for your help,