fixed a long lived bug in do_forces. Rrf was not being used in the neighborlist correctly. rcut was conssistently being set lowere than Rrf causing the dipole cutoff region to be to small. Also led to the removal of the taper region to buffer the dipole cutoff.