Meta analysis are held to much greater scruple than RCTs.
63 is a very low number when, as that PLOS article reports are in the hundreds.
I admit to having a bit of difficulty with parsing your English, but
1) the article doesn't really mention the amount of studies in the average meta-analysis. So you can't deduce from it that 63 is low.
2) just that the majority of meta-analyses are underpowered to indicate a large effect, but
2a) This atheism meta analysis does not indicate a large effect
2b) underpowered does not at all mean it's wrong
2bii) in fact the authors mention themselves that the studies might actually be fully adequate for more specific effects
2c) the authors of the meta-meta-analysis included reviews with at least five studies. I can tell you five studies is almost per definition underpowered
2d) almost, because pooled N tells you much more than the number of studies
2e) given that from a cursory glance a lot of the reviews are about small scale interventions for specific problems, it's guaranteed that these reviews have an overrepresentation of small scale home studies, that's much less apparent in the atheism study.
2f) as said giving this kind of criticism would require a more in depth of the original studies in the atheism meta analysis. I'm however much more inclined to trust the reviewers of one of the most esteemed journals in the field, than a casual dismissal from random forumite with a study from a different field that at best casts some doubts over strength but not much more.