Often the difference (if there is one) is just dynamic range. Sometimes it might also be EQ or less likely something else, none of which can’t be achieved audibly transparently with 16/44.The difference is not just the dynamic range, so you maybe should not focus too much on that. Its the overall sound that is different.
Because anything higher than that makes no audible difference. However, it does make a difference to the marketing and to the price you can charge if you can fool audiophiles into believing that marketing!But beside that, why would you release the audiophile targeted Version in 16/44?
And how exactly can you “just export it”? Regardless of what bit depth you record at, all professional DAWs mix/master at 64bit float and you cannot export a 64bit float as wav or any other common audio file format because they don’t support it. In fact no professional DAW or digital mixer as far as I’m aware has ever mixed at 24bit. So there is no choice but to “take the additional step and convert it”! The only potential question is what to convert it to.Why take the additional step and convert it? If you record in 24/96, it makes much more sense to just export it and be done.
Much of the rest of your post is partially incorrect but in any case, all of it could be covered by having two different 16/44 versions, say a standard and a high dynamic range version.
G