In my case, what tends to happen is that I either don't notice the question, or I notice that the question requires a bunch of work to respond to and then either get to it some time later or let it slide off entirely.

I think in cases like these, you're more likely to get a response by adding another post as a reply to the person with just a single unanswered question (start with the one you care about most), so the person will see they have a new response in their inbox and realize they never answered an earlier question. If you post each of those 3 questions as a response to the person, in context, I'd be very surprised if you didn't get a response to at least 2 of the 3, as long as you include little to nothing else in each post so it's obvious what you're asking for and they can't respond to something else in the post.

I've noticed that longish posts with multiple questions often get just one question answered and all the others ignored, intentionally or unintentionally. And posts that are longish with questions interspersed with non-questions tend to get responded too as if the non-questions were the substantive part, with the questions often ignored.

(The other extremely common reason for not getting a response is identifying a flaw or asking a question that shows problems with the person's position, in which case most people seem to just ignore the post rather than admit they were wrong or can't answer a critique. I don't think that's the case here at all though.)

Not answering a question is Internet's way of walking away from a conversation. You don't usually say "excuse me, I'm late to a meeting", as interaction is asynchronous. In the current Internet culture, saying "I don't want to bother answering" sounds rude, and so the best solution signaling-wise is to just not answer.

The first two have responses as of the time that you posted this. (If there are remaining questions in a post they are far less likely to be answered after the first couple of replies.)

The final example suffers somewhat from 'nobody knows what science doesn't know'. There are probably not too many people who can think of an example of a problem that UDT1 can not handle. For my part I probably wouldn't answer just because I don't like the name UDT1 and the language used to describe it irritates me.

I'm not sure why Eleizer didn't answer but I probably wouldn't bother wasting thoughts wondering. Want an answer? Make a top level post about it. Include enough of a useful description of the theory and the problems it has already solved to make you not look bad. In particular, include links to said problems and resolutions. If you really want an answer then include an assertion that UDT1 has solved all the significant decision Problems that have been discussed on LessWrong.