But in this research, the intent is to let the embryos develop into full human beings. It would be unethical to edit such a genome if it is uncertain whether or not the individual will suffer due to these changes.
The problem they talk about is that you trade one deadly or very damaging mutation with a couple dozen other mutations with unknown consequences. These are almost certainly harmless, but the risk of introducing another catastrophic defect is still too big.
To make this tradeoff more secure you would need to sequence the embryo's genome at a certain state in its development (after gene editing, but before transplantation) and check it for errors. But we neither have the technology to get accurate genomes from a few (or just one) cell, nor do we have the genetic maps required to tell if mutations have consequences or not.