This is an extremely unusual conception of epistemology, even for mathematics. This knowledge that is contingent on axioms isn't at all convincing. At the begin

This is an extremely unusual conception of epistemology, even for mathematics. This knowledge that is contingent on axioms isn't at all convincing. At the beginning of the 20th century, this was a raging debate. It wasn't quite resolved, it just didn't quite matter to practicing mathematicians so it kind of faded into the background. There is a massive gap between 3 and the cardinality of the continuum. 3 is directly examinable. If I count some collection of objects, my fingers say, I will immediately grasp 3. On the other hand, the set of real numbers is a highly pathological, abstract concept. Everything we know about the reals suffers from two severe deficiencies: One, it depends on infinitary axioms which presuppose facts about the phenomenon we would like to investigate. Two, even what we can infer from such axioms is always indirect evidence. That's not surprising: All reasoning about infinity is indirect. In certain mathematical settings it is even true that the Cauchy sequence definition of real numbers and the Dedekind cut definition do not agree. At the very least, the reals isn't even a thing: there are many reals. That you say that "we know the cardinality of the reals" because we know CH is independent of ZFC is... preposterous to say the least. "If you assume you know the cardinality of the reals, then you know the cardinality of the reals; therefore we know the cardinality of the reals" is basically what such axiomatic acrobatics boils down to. Axioms are not knowledge.