You're asking good questions, but it's unreasonable to expect one paper to answer all of them. Probably the article should have emphasized more strongly that mouse models are imperfect, but they do show efficacy in two different mouse models, which counts for something.

> It fails to prove that fixing NAD+ in humans will stop the disease, only that it works in mice engineered to have the disease

This in particular is just not possible without clinical trials in humans. But you can't have a clinical trial without evidence of efficiency, so you need to start with mouse models, even if they are imperfect. Sadly we don't know if any of the existing mouse models are any good, but it's the best we've got.