By choosing to treat non-foundational issues in a single unified way that is distinct from how we treat foundational issues, we keep our thought more ordered and localize whatever problems there might be to our axioms.

That just says they are different. They have to be, because we can pin non-foundational issues to foundationail issues, but we can't pin foundational issues to foundational issues. However a difference is not the difference* -- the differnce tha would show that any arbitrariness of foundations affects what is founded on them

If [empiricism] turned out to be false (say, if we were programmed from birth with many innate truths), we could still do science

I suppose there could be a weak empiricism that just fills out the gaps in apriopri reasoning. However, it is doubtful that apriori reasoning can supply truth at all. See below.

It's also worth noting that the logically knowable truths are far greater in number than the empirically knowable ones.

So long as you are willing to accept valid derivations from arbitrary premises as actually true. One can derive all sorts of things from the cheesiness of the Moon..