But then, by your own criteria, any explanation of that concrete, observed fact, would be "metaphysical". IOW all of science other than "concrete facts," is "metaphysical".
What you're saying now is that "drawing conlcusions from *facts* is metaphysical because it involved 'abstract thought'". Therefore it doesn't matter whether the subject of a theory -- the phenomena being explained -- occurred in the "unobserved past" or in the "observable, repeatable" present.
So I take it you're now abandoning this "unobservable past" criteria?
Nope, only if the purported events are limited to the unobservable past does the metaphysical part of the definition come into play.