The case of black holes is illustrative of the conclusion I am going to offer, so I'm going to dwell on this case to some extent. Now, one might construe our knowledge of black holes as having been a priori in the sense that, before we ever "saw" one, we predicted their existence because the possibility of this existence was as a possible solution to certain deep equations that we accepted (albeit on empirical grounds, so to speak). One interlocutor in such a dialectic might object that making a true prediction is not tantamount to advance knowledge, due to the epistemic nature of predictions. They might quibble that we had "preknowledge," but not knowledge strictly, of black holes, up until recently.
The next counter might be that, besides knowing of black holes' general possibility via the relevant equation-solution, we also were aware of specific dynamics, in our world, that might form them, i.e. stellar evolution over sufficiently massive stars. But then a fringe (hopefully not pseudoscience) researcher might come along and say that our current more "direct" black-hole knowledge is consistent with the possibility of eternally collapsing magnetic objects, and appeal to the alleged theory-ladenness of observational evidence to defend the suggestion philosophically. Similarly, though the prevailing model of active galactic nuclei is as a function of supermassive black holes, there are outliers who ask after boson stars as an alternative model. And then an ecumenicist in physics, so to speak, might go on to propose that black holes are boson stars made of gravitons; or that black holes and other types of boson stars are both possible and might be the AGNs of different types of galaxies.
So for all that, how much apriority was involved in knowledge/preknowledge of black holes? All the "pure reasoning" in defense of these objects was not so pure, after all, requiring empirically-supported priors of various kinds (e.g., again, our largely empirical understanding of stellar dynamics).
But just the same, then, since there was apriority as epistemic proactivity involved, it ends up looking like Kant was on to something in trying to harmonize the motives of rationalism and empiricism. Since "the point generalizes," we might find ourselves thinking that very little is known in total abstraction or entirely due to physical experience ("Necessarily X means not possibly not X," as a representative example on one end, "I am having red-gradient color perceptions right now," on the other).
For technical reasons, I do think ethical knowledge, such as it is (or is supposed to be), would have to be heavily a priori, its empirical supports limited pretty much just to immediate perceptions, on pain of having to derive action-guiding principles from epistemically difficult claims. The old (discredited) notion that the fact of evolution justifies Social Darwinism hearkens to a commandment that Kant said was analytic, viz. "Act according to the truth." So the Social Darwinist would reason that contrary moral perspectives involved acting against the truth of evolution; and we find such specious thoughts in many a moral theorist's repertoire. Whence the speciousness, here? Just in that, if local moral imperatives are connected to universal, eternal moral facts, we'd best not have to learn the latter only after fairly complex observation and research: or else, with respect to evolution (say), we would be asserting that people were obligated to "act according to the truth of evolution" (or whatever) even when unaware of the process. (This is of a piece with the mainline objection to standard formulations of utilitarianism requiring us to perform overly elaborate future-oriented calculations of universal utility functions, which are not practical for us to perform.) But this leaves us with having to be able to recognize moral information a priori; yet if moral principles involve a proactive capacity inside us, and if apriority just is epistemic proactivity, what we are left with is something almost "true by definition" indeed (hence Kant's derivation, in the second Critique, of the categorical imperative).