I agree there are large classes of things like that, but I think there are also large classes of things where looking them up is fine--- and my guess, though I could be wrong, is that the things traditionally memorized in schools lean more towards the second. If you think of the kinds of information that comes in tables, I think it's important to know that the table exists, and what kind of information is in it, when you would want to consult it, and ideally even how you would recreate it if you needed to. But actually memorizing the table? Doesn't seem that useful, and I think that's the kind of thing traditional tests have focused on (say, memorizing a bunch of different combinatorial identities, or memorizing a bunch of properties of chemical elements).

That's especially the case when nobody in the real world would be in some bizarre situation where they're stuck on a desert island having to do things entirely from memory. For example, if you're doing big-O analysis of an algorithm, it's perfectly reasonable to assume that you'll have Maple or Mathematica, or at least a list of common identities, available to use when simplifying your result or solving recurrences.