This PR changes the invalid mode approach to get mixed scalar/vector problems working in spherical coordinates. In spherical coordinates, Hermitian symmetry is only imposed for ell=0 modes for scalar and vector fields. This seems to fix all issues with various boundary conditions and gauges in scalar and vector problems. Issues may still arise with BCs with tensor fields in spheres, as well as in polar coordinates, where Hermitian symmetry will need to be imposed using real projectors as preconditioners, rather than just by dropping modes.
This PR changes the invalid mode approach to get mixed scalar/vector problems working in spherical coordinates. In spherical coordinates, Hermitian symmetry is only imposed for ell=0 modes for scalar and vector fields. This seems to fix all issues with various boundary conditions and gauges in scalar and vector problems. Issues may still arise with BCs with tensor fields in spheres, as well as in polar coordinates, where Hermitian symmetry will need to be imposed using real projectors as preconditioners, rather than just by dropping modes.