As far as I can see the basis set you are using contains linear dependencies. The code removes those and zeroes the linearly dependent vectors. It seems that the routine that purifies the symmetry of the orbitals gets confused about these zero vectors. I am having a closer look to see where exactly this happens and fix this.