Now that I see your answer I have a better idea of your question, and would like to attempt a slightly different interpretation.
I do not have the right editions to cite, but am basically looking at Preface to CPR,2E (very helpful), Third Antinomy, and Canon of Pure Reason, Sec. I.
I take it that Kant's main concern is not only Hume's skepticism, but the broader issue of his day: How can a natural causal determinism (the basis of scientific knowledge) be reconciled with freedom (the prerequisite of Protestant morality duty or "practical reason")? He is most concerned with preserving "practical reason" against skepticism, utilitarianism, determinism, atheism, and dogmatic authority. Even if sure knowledge loses out.
First, as to the second part of your question. I think he takes our freedom "in experience" as unproblematic.We can simply observe ourselves as "entities" among phenomena and see that we choose this or that. Moreover, to engage in science and discover "causalities" (efficient or otherwise) also requires speculative freedom. We can insert "spontaneity" into the causal chains to do "experiments." Notably he calls this not some "absolute freedom," but a type of "second causality" to preserve, I believe, its moral imperatives. So our "freedom" is traceable as an uncaused, spontaneous "cause" among sensible things, but not determined by them. Hence moral duties.
Now the difficulty. This tells us nothing about the origins, limits, or demands of this practical, experiential freedom. It is free of efficient "causality" and must therefore originate in the noumenal or transcendental realm. Here it can enjoy "logical relations" apart from temporally "causal relations." Like everything else the will or "soul" has a double existence as phenomenal and as "ding an sich." The problem is, how does Kant know? We presumably have no access to such noumenal entities.
In the Preface, 2E Kant argues that we cannot "know" we have this freedom. We cannot "know" things-in-themselves, such as our soul. But we can still "think" them hypothetically, as long as they are not self-contradicting. So we can deduce this freedom from (1) the observation of experience, (2) the fact that it is not rationally contradicted, and (3) the necessity of some transcendental origin of moral freedom. Such knowledge enables Kant to do what he really wants to do. Which is to makes rational claims about this freedom and practical reason, preserving us from epistemological and moral anarchy.
My understanding is that Kant rewrote this material several times and that nobody, himself included, is entirely satisfied with it. To have his "two causalities" he must separate the phenomenal and noumenal, but must then tell us how he can make claims about the "noumenal" origins of "practical" reason. While I agree with the substance of your criticisms, I am not sure they affect Kant's overall argument. Science can always gain more and more knowledge of cosmological or neural "causes" in the sensible realm. But we can never secure any "final" causes except with regard to moral-practical reasoning.