This specific technology wouldn't be applicable to designer traits, as it doesn't modify the nuclear DNA in any way.
Crash course on mitochondrial genetics
There are approximately 23.000 genes in the human nucleus, coding for nearly every process in the body. The nucleus doesn't hold all the DNA of the cell though. Billions of years ago, one early life form absorbed another, and they began a symbiotic relationship. The absorber provided shelter and nutrients to the absorbee, who specialized in the production of energy. As time passed, the absorbee gradually ceased to function as an independent cell, instead acting as what we would now call an organelle, the mitochondria. The mitochondria's fitness was no longer influenced by genes that did things like build protective structures, so mutations went unchecked. Over time, these genes ceased to function completely.
The human mitochondria of today has only 37 genes, most of which are involved in its own processes. Because these genes control functions of energy production, they are VERY important; defects in some can have catastrophic effects on the cell, leading to eventual death. For this reason, the mitochondrial genes vary by very little over time; small changes can have very strong repercussions. (Answer in advance: differences in mtDNA between individuals is very largely centered on the non protein coding regions, to which you might have heard referenced as "junk DNA.")
Mutations in the DNA do occur though, and often lead to a collection of disorders that we call "mitochondrial diseases." Not all mitochondrial diseases are caused by mutations in the mtDNA; some aspects of energy production are controlled by nuclear DNA. This therapy would not be helpful for nuclear coded mitochondrial diseases, so they will be excluded from future reference.
Mitochondrial diseases have a wide range of symptoms, owing largely to the presence of mitochondria in nearly all eukaryotic cells. Essentially, what goes wrong depends on where the defective mitochondria are located, with some regions being preferred over others. Muscles and neurons tend to be the most strongly affected, leading to, respectively, fatigue, weakness, paralysis, and pain for muscle tissue, and nervous dysfunction (ranging from dementia to blindness) for neurons. The progression can be variable, but most mitochondrial diseases kill either in infancy/early childhood, or in early adulthood.
How do these defective mitochondria continue to propagate? Essentially, most cells in your body have a good number of mitochondria, ranging from less than 100 to more than 10.000. One defective mitochondria won't ruin the lot of the cell; the mitochondria won't operate properly, which might have some impact, but there are others there that pick up the slack. Most mitochondrial disease clusters have a threshold level of necessary defective mitochondria to display clinical symptoms, that is there's a tipping point, after which the cell will be unable to operate properly. I don't know many off the top of my head, but the Kearns-Sayre/CPEO cluster tends to be about a 65% defect threshold.
Replacing the mitochondria of the embryo's cells will only affect the reproduction of future mitochondria; it won't have an effect on the nuclear DNA. The treatment offers a promising step towards treating mitochondrial diseases, but in its current state it has no use in allowing gays to have children, or allowing parents to design the perfect child.
