Only talking about the reason I gave for it not working, most restriction enyzmyes recognize 4-6 base pair sequences of dna. Therefore, any such restriction site in the genomic dna would be cleaved (not just myostatin). We're not even talking about the other million problems.
Honestly, I think gene therapy is possible, not through the methods you suggest but rather through RNAi or the like.
Right now the "best" way if someone were to try to inhibit myostatin function would be to use neutralizing antibodies. Based on mouse models, if I remember correctly this would equate to a ridiculous dose of like 5g/d.