Not using the multipole expansion first will make the math very very challenging in most cases I think, if not absolutely unsolvable.
1) I think the multipole expansion is a valid approximation if in the limit it converges to the actual answer.
2) For perturbation theory you do need to consider degeneracy in quantum mechanics but besides that, I think it is the same answer as for the multipole expansion. Besides that, it is important that the perturbation remains small.
3) I do think it is feasible to realize these conditions simultaneously.