Wrong list 4 cost estimate

In this line in the cost estimates for list 4 we have:

        result["form_local"] = nform_local * xlat_cost(p_fmm, p_fmm)

However the cost of forming a local expansion due to a point is \Theta(p_{fmm}). This could be the reason we are seeing unreasonably high list 4 cost estimates in the model.