Ordinary natural selection, including human sexual selection, works this way too. I still fail to see why you worry.
Say a couple want children, but they want a red-haired girl, even though their genetic combinations make that an impossibility (just for the sake of an example I took something from the 'cosmetics' department, not from the disease department) you could conceivably splice in the genes to make that happen.
The problem here is that even though the resulting phenotype may be the desired one that there could be any number of interactions between the 'undesired' genes and the rest of the genome to keep the genome as a whole stable across generations.
It will take a long time before we will know if that's the case or not.
Why would you want that, if every generation has the option of designing their children anyway? And they can always revert to a backup genotype from 2000 if there's no other recourse.
I'm afraid you've got me there, this is speculative in nature, only time will tell.
Let me try another angle then, to me evolution is not about individuals, individuals are merely the current vessels holding a single copy (with possibly some local variations if you want to be very precise) of the status quo in contemporary genetics as perceived by that individual.
Sexual reproduction is in the majority of the cases the way in which this genetic material gets mixed and slowly, over generations these changes + mutations introducing new variation have converged on the diversity that you can observe around you.
To find the 'keys to the kingdom', to get 'root' on this system is of course a very powerful tool, but if you don't know exactly how powerful a tool is or how it works you would be wise to meddle with it as little as possible until you do have that knowledge.
Keep in mind that we will be selecting for phenotype only, what's going on under the hood is a question that your great-great-grandchildren might be able to answer, and if our luck of the draw runs against us it is possible that we introduce some effect that will only become apparent to them, we'll be long dead by then.
Those effects could be worse than the things we are trying to cure today, and I'm merely cautioning to be careful about unintended side effects of over enthusiastic modifications to our 'imperfect' genome.
It has served us well, let's give it the benefit of the doubt until we have solid proof that we are not going to create a larger problem.