That's a good argument for not engineering all children all at once - we don't understand the consequences. It seems like a good argument for starting to engineer a few children so we can start to learn what the consequences are. As engineering becomes more palatable to the general population we will better able to deliver good results.
Engineering children would likely introduce horrible diseases because as you say, nobody understands the consequences. What's being proposed is selection, the selection of embryos which have already proven themselves to be viable and passed several steps of Nature's disease filtration.
