The first 1/3 is sort of useful, I'd say stop reading the deck when it says “There are invisible, undetectable elves all over this room” is a meaningless stence (it is not meaningless, the sentence may be false but everyone can understand what it takes for it to become true, namely 1. you see elves in this room OR 2. you detect elves in this room OR 3. both.).
Also, I'd suggest epistemiology is not so much about finding out what's true or false but what is knowable and not knowable in principle - i.e., to establish the frontier of the knowable (which does not change, whereas the frontier of the present knowledge shifts).
No, your 1, 2 or 3 don't help at all. If you see the elves or detect them, then they are not "invisible, undetectable elves", and your observation has no bearing on the truth of the sentence. So the sentence absolutely cannot ever be tested, cannot be said to be true or false, and therefore is in some sense outside the domain of logic.
You can have a separate argument about whether an untestable statement is necessarily "meaningless" (maybe the way it makes you feel is the meaning) but I believe the only point the author is trying to get across here is that some statements make predictions about the world, and some don't, and it's worth being aware of the difference.
Also, I'd suggest epistemiology is not so much about finding out what's true or false but what is knowable and not knowable in principle - i.e., to establish the frontier of the knowable (which does not change, whereas the frontier of the present knowledge shifts).