Efficiency and accuracy issues aside, I think there's one case and only one case where editing embryos is acceptable: correcting a thoroughly understood disease mutation back to the human genome reference sequence. This is the only case where we know as close to exactly as possible both how the edited genome will influence the embryo (because it's being changed to a sequence that's already vetted in almost every living human) and that the resulting edit is unambiguously beneficial. Anything else is essentially an experiment on an unwitting human subject.

If there's ambiguity in either the reference sequence (like targeting the regions of the genome responsible for individual diversity) or in the "benefits" conferred by the edit (which would be the case for frivolous cosmetic changes), that edit isn't just wrong, it's completely unconscionable.

If an edited embryo is allowed to develop, it's going to produce a conscious individual that has to live with modifications to which it did not consent. So, conferring any edits which aren't so obviously beneficial that nobody could rationally not consent is a violation of the right to self-determination that we should all strive towards, if not have already.

Edits that benefit the parents or society more than the individual are absolutely inappropriate, as are edits whose effects are not completely known (and thus carry risks that the edited individual didn't consent to.) Thus, I think the insertion of novel DNA is and never will be acceptable.