We would need a very particular set of conditions for embryonic editing to be justifiable under a medical dogma that aims to "do no harm." Both parents would need to carry a large common set of recessive deleterious alleles, as this would make embryonic selection of non-carriers very difficult. Then we would need the editing system to be so reliable as to not introduce off-target mutations. In a preimplantation setting, we can't easily observe if non-desired mutations have been introduced in some cells, as this would require sequencing every cell in the developing embryo. Serious disease introduced through chimeric errors in the editing process would be a real possibility, and there is no feasible way we could guard against this result using sequencing as it would require destruction of the embryo.

A more realistic scenario would be to develop a human embryonic stem cell culture that has been edited as desired and then implant this into a developing blastocyst at a point at which it would take over the and develop into the fetus. This is done with mice and there is no reason it wouldn't work for humans. I think that most people would find this much more abhorrent than directly editing the germline. However, it would be much safer for the engineered proband and would not require a "perfect" editing system that we do not have.