Helpful reply! But two things make me uncomfortable with this kind of meta-study in addition to the "no mechanism" part.

1. Short term effects, why do they matter? Is physiologically or psychologically masking depression an interesting or useful thing? Maybe the former, probably not the later. That's partially what my bee sting thing was about. I'm sure there are many ways to mask depression, and the weaker your a priori mechanistic justification needs to be, the longer or more ridiculous the list will be.
2. Meta-studies like this one might really suffer from publication bias. They aren't sampling a random pool of studies. Rather, because positive results are more likely to be published, they are more likely to be included in this study. Therefore, such a meta-study would overestimate any effect, even more so than most individual papers. I notice only a few of the studies they sampled had effect sizes that overlapped with zero (Fig. 2 below). Including only randomized samples could help with this!

Just some thoughts! Criticizing is way easier than building up and the authors did impressive work.