The genetic mix from parents is randomized. Always. Both good and bad results are equally the result of a random process initiated by the parents. A combination of recessive genes to cause disease is not more or less random than a combination of healthy genes. The parents roll the dice, the child gets the result.
In other words: The ethics can't proceed from first principles around consent to experimentation. It has to include, as you say, empirical questions about the technology in question. But that's a different sub-thread.
If I had to say anything about that: I think this is one of those things that is just going to happen whether or not anyone likes it. The fact that it has already succeeded with mice + the massive applications... it's just going to get tried, it's just a matter of which country it happens in. The fact that it could potentially be used by same sex couples is a total side note to how strongly this will be pulled along by the same existing demands that drive IVF currently, the potential is just too powerful. So I imagine we're just going to see eventually.