I would want to add the a priori that we humans experience things that are not satisfactorily explained by a humanist or natural metanarrative.  Miracles, for example, are not simply mythological artifacts but the preferred explanation for things people experience in the 21st century.  And not only naive, ignorant believers experience such things, but also sceptical and hostile adherents of other metanarratives.

I am not saying that there aren’t natural explanations offered (frequently having to do with psychology, faulty observations, or ignorance that could some day be remedied when science has advanced further), just that these explanations are rejected in favor of competing "unscientific" ones.  Where the question of miracles is investigated scientifically (and this happens far more infrequently than science lets on), the results are frequently inconclusive (typically with appeals to possible psychological causes, methodological questions, or hopes that some rational naturalistic theory may come along to explain everything).

It would no doubt be dangerous and unpopular, but I for one would welcome rigorous scientific testing of religions.  Studies have been done, for example, demonstrating a statistically higher recovery rate from serious medical conditions when the patients are prayed for; I’d like to see how this breaks down by religion, gender, denomination, etc.

Certainly many things are beyond the ability of Science to repeat or observe (e.g. cosmogony, ancient history, people’s thoughts) but there remains much about modern religious experience that should be repeatable and observable, at least over large populations.  Once Science has given this serious attention, it may be more appropriate to discuss the extent to which "religion" (or some specific variant/s) is a human construct.  Until then, I’m just hearing "Religion is unscientific because there’s no such thing as God" or "There’s no such thing as God because there’s only nature"—both a priori presuppositions.