Why did Hume think that one could justifiably believe that an extraordinary event had occurred, under certain circumstances, but that one could never justifiably believe a miracle had occurred? The proposed interpretation of Hume's analysis of miracles in relation to his analysis of causation and his wider empiricism yields the only plausible answer to this question that I know of. This interpretation also shows why it makes no substantial difference whether we interpret Hume's argument in Part I "Of Miracles" against the possibility of justified belief in testimony to the miraculous as an a priori argument or an a posteriori argument since the arguments essentially coalesce.
My own interpretation is that there is no argument at all against the possibility of justified belief in testimony to the miraculous in Part I of the essay. Rather, what Hume does is argue for a standard of proof to which testimony for miracles would have to be held. He then goes on in Part II to argue that no testimony in favor of miracles meets this standard.
The only basis, I think, for holding that there is an argument against the possibility of justified belief in testimony to the miraculous is the principle of counterpoise. Hume says:
The very same principle of experience, which gives us a certain degree of assurance in the testimony of witnesses, gives us also, in this case [i.e., the case of a fact "as has seldom fallen under our observation"], another degree of assurance against the fact, which they endeavour to establish; from which contradiction there necessarily arises a counterpoize, and mutual destruction of belief and authority.
It is the principle of counterpoise that seems to make Hume's argument parallel to Tillotson's argument against transubstantiation. Thus one might argue, that just as Tillotson uses a sort of counterpoise-principle to argue against transubstantiation, Hume uses it to argue against miracles.
This can be granted; but it doesn't follow from this that it has any teeth in Part I. (It clearly does in Part II, because if the considerations in Part II are right, the mutual destruction of belief and authority always ends against miracles.) What Hume actually argues in Part I is that even if the testimonial proof of miracles be 'entire' -- i.e., conclusive if considered on its own -- in combination with experience of the uniformity of nature (also a complete proof in Hume's sense of the term) it creates a counterpoise, to the mutual destruction of both. If there is any disparity at all in the force of the proofs, then the proof with greater force wins out, and survives with its original force minus the force of the opposing proof.
Given that Hume had originally told us in Section VI of the Enquiry that proofs are arguments from experience that leave no room for doubt or opposition, it is something of an exegetical mystery how Hume can talk about proofs in opposition, but he does; so either he is inconsistent or (perhaps more likely) he is merely speaking hypothetically here or (and this, while more of a stretch, is an intriguing hypothesis) he is describing a mechanism for how proofs could cease to be proofs. In any case, I won't dwell on this here. The question at hand is: Does this amount to an argument against the possibility of justified belief in testimony to the miraculous?
No, unless one assumes that testimony for the miraculous never has a greater force of proof than experience of the uniformity of nature. Hume never argues for this in Part I. Indeed, strictly speaking, he never argues for this at all, because he thinks it can easily be shown that testimony for the miraculous doesn't even have the force of proof (unlike experiential evidence for the uniformity of nature, it always falls short of proof). Strictly speaking, all Hume implies in Part I is that you'd have to consider the matter on a case-by-case basis, and determine whether (in his words) it would be more miraculous for the testimony to be false than for the alleged event to occur. This is not an argument against justified belief in testimony to the miraculous; it's the set-up for an argument to such an effect, an argument we only get in Part II.
I like parts of Levine's argument, but I think in addition that he puts too much emphasis on the fact that in Hume's sense we could never experience a supernatural cause. Hume seems elsewhere pretty well committed to the claim that we could never experience the cause of gravity, either; but it doesn't follow that we can't ever attribute events to it, due to an inference that there must be such a cause. (This particular example gets into tricky issues in Hume interpretation, but Hume does clearly allow for us to attribute events to causes we can't experience; he attributes such actions to 'philosophical', i.e., scientific, thinkers.) Also, I don't think Levine's quite right about the Indian prince, although I find his argument very interesting. The Indian prince example, like the Cato example, is just an example of counterpoise, as the way of proportioning belief to evidence, in action. Due to Locke, this sort of example is in fact used in this period as a rather standard example of such a proportioning. (I've briefly discussed it here and here.) And note, that while Hume insists that the Indian prince reasons justly, he denies that the Indian prince is in a position to be "reasonably positive" about what would happen under conditions other than those found in Sumatra; the claim that ice freezes requires "a pretty strong testimony, to render it credible to people in a warm climate". This seems to me to be fairly conclusive evidence against Levine's interesting conjecture that Hume considers us all to be in the position of the Indian prince (with regard to testimony for miracles). Hume actually denies that with respect to miracles we are in the same position the Indian prince was with regard to ice. Hume says something very interesting in this connection, though. He says that the difference between the two cases is that the Indian prince actually doesn't have a proof against ice freezing under different conditions (it would be a "new experience" and thus uncertain, and liable only to analogical conjecture; and analogy would have not have suggested the freezing of ice). This differs from the case of miracles because the miracle is supposed to be a deviation under the same conditions. This is why the experience of the uniformity of nature is supposed to be a full proof against a miracle we are told of, even if we are supposed to have a full proof in favor of the testimony by which we heard of it.
But these kinds of issues are a bit contentious. I like Levine's article, although I disagree with much of it. I do agree with his claim about Bayesian interpretations of the argument; but I think my interpretation gives a better reason to accept such a claim than his own does.