But you are not taking into consideration 2 factors. Parents want a limited number of children (with exceptions). Parents for all intents and purposes have unlimited embryos (only limited by the number of ova the woman has). If sequencing can come down to under 100$ (who's to say it won't be next to free) then you supply 200 ova, enough sperm and you develop embryos in the lab.
Now when they have become a zygote you DNA sequence the lot to find the best (with least flaws/potential vulnerabilities) and then implant enough of them to get 1-2 successful children and the rest are discarded.
When it becomes wholesale, instead of on a case by case basis, it will become the norm and what is ethical will change to fall in line.
I agree that this is both the most logical and most likely form of using genetics on embryos. However, I see a problem with the massive selective pressure this will create. As you point out we already do this to some degree, but the degree matters. Genes are complex things, and while I have no doubt we will one day have a very thorough understanding, that day will likely lag behind the ability to screen embryos.
What happens when a gene that increases risk of Alzheimer's disease also confers other positive traits? No parent will take the chance on that gene, and it will disappear in a generation. The same will happen to many genes with any possible negative qualities. The result will be a much less genetically diverse population. As I already pointed out, many genes that will be wiped out will also have positive traits associated with them. By only selecting the 'best' embryos the human race could end up much worse off.