How about a hypothetical situation:
Dark night. The aircraft is not loaded symmetrically, so it WANTS to roll. We loose all gyros. We have a glider style slip indicator (a string on the centerline of the windshield, in the slipstream) and we religiously keep the string centered with rudder. So no sideslip. No lateral inputs (since we have no gyros). Are we going into a spiral, or does it depend on our anhedral / dihedral setup?
(BTW: I don't think pitch matters here.)
So I think I agree that all roll disturbance recovery and final stable equilibrium will involve some sideslip. I don't know whether this helped the OP, but it got me straightened out a bit. Thanks.
And to answer my own prior dilemma about 1st / 2nd order influence of relative position of center of lift w.r.t. CG: it must be 2nd order. The moment of unbalanced lift (in a slip) must be a much larger effect, considering typical wing spans.