If that's true, how come science and scientists rely upon a priori theories and make claims about things which have never been observed and cannot possibly be observed, or in places where there is no observer?
There is a difference between "can be observed" and "actually have been observed." Science is free to hypothesize about forces they have not observed, the constraint is that their hypothesis be falsifiable. Supernatural explanations are not.
Their hypothesis is not a theory until evidence supporting it arises, however indirect proof is enough. If I have a theory that tells me how oil is formed, and based on my theory I can also make predictions about where oil will be found, if my predicted locations p[rove to be better than random guessing on a statistically significant basis, that is evidence for my theory, even if I never see the actual oil form or an actual instance of evolutionary speciation.
"In physics as ordinarily set forth, there is much that is unverifiable: there are hypotheses as to (a) how things would appear to a spectator in a place where, as it happens, there is no spectator; (b) how things would appear to a spectator in a place when, in fact, they are not appearing to anyone; (c) things which never appear at all." -- Bertrand Russell, Our Knowledge of the External World
And those things, or the explanations of them are not science, they tend to be metaphysics (which is still not the same as religion). What happens to an electron when no one is observing it is a good example, as the answer is "we do not know." There are interpretations of quantum mechanics that purport to give an answer, but no one confuses those interpretations as scientific theories. That's why part of the Copenhagen interpretation can be boiled down to "Do not ask those questions."
Scientists sometimes do create thought experiments of the sort Russell notes, but usually as illustrative aids or as a result of speaking loosely rather than because the scientists believe they "know" how such situations would really be.
In other words, it imitates science.
Again, science works with theories that are falsifiable, and in order for that to be the case, it has to be possible that when turning to nature (or experiment) you will find evidence that cannot be explained by your theory. Even if you *never* find such contrary evidence, it has to be possible, in principle, that such evidence exist or else what you have is not a scientific theory, but a metaphysical theory.
In religion, the basic assumptions are non-falsifiable. My omnipotent God took Action
X. There's no evidence that disprove the acts of an omnipotent God. Even if you can explain the same thing using natural forces alone, perhaps God make the natural forces operate. Any argument offered a believer can dismiss as you are left at bottom with the equivalent of the trying to prove a negative: trying to prove that God does not exist or could not have done a thing.
*Some* metaphysics is religion, but not all of it. Religion tends to rely on agencies that are separate from natural forces (i.e. supernatural forces) and able to vary or ignore the natural forces to accomplish their ends. Religions also tend to have a broader scope, in that they try to explain significant parts of society or the world.