(But I don't want the poll to be on FB itself since many people wouldn't be able to use it)
It's not that we can't. We just know better.
The use of a dedicated tool is a must-have
Eventually, yes certainly. This really wouldn't be too dificult to develop I would of thought. Personally with regards to "additional services" I'd be developing a portal interface first. Somewhat like a "kiosk interface" granting access to various tools which are integrated to the maximum way that still makes sense. Then you can attempt to build individual things in a modular framework around this portal. Then we might end up with something that can do anything, and is still actually possible to maintain or upgrade.
But for what we have made available right now - ie: a forum - Then a subsection can hold a months questions or so at a time - each question can be additionally a poll, if it'll allow one option then we can get a good handle on how many are interested - sort to the higher polled first - and anyone with similar or additional of ilk can in theory tag into that. Over the month, mods can keep it trimmed(taking on extra for that purpose if required) and tidied. A "dead question" subsection can exist to simply throw all the posts into after processing(long term retention in DB). This should make it easy to total them all up into a concise set of questions over a few days and submit for answering. I'd think it decent to attempt to give him a reasonable window in order to come up with answers, I suspect he will have cause to think before answering a few. That pause will add up over 20-50 questions. To assume Dr Ashurebyli's translator costs him he can deliver his response in his native language and surely there will be those in our number able to translate into every relevant language - reducing time and cost to him further. A subsection next to the questions subsection could potentially contain responses archived over time - I'd personally think a wiki more suitable, I can't recall if that feature request was picked up and confirmed inbound tbh.
External tools - especially things of closed source nature, or of proven privacy and or security risk(like facebook, and you can add in moral and ethically questionable conduct with regards to the data they rape from their victims, I mean users.) are definitely things to avoid. It's honestly better to not even bother starting that to intentionally build on shakey foundations. It never ends well. To ignore any privacy or security concerns, I've seen too many things shut down and isolated purely because they had no control over the medium they attempted to utilise to consider trusting random third parties anywhere in the chain, no matter how trivial the data.
Initially, there is expected to be a high volume of questions - but retaining and displaying previous sessions should in theory minimise this over time and eventually those tasked with pruning questions in there will be mostly consumed with checking it's not been previously asked(we need something on the post input system to check for such - maybe not prevent post outright, unless clear inhuman posting demonstrated, but flag for moderator attention. This could be adapted to catch spam runs etc additionally (another reason for portal, there's so many tools...) but this could be achieved with a little script in cron throwing the ngnix logs through awk producing a "report" of things what might want to be looked at. For post count sorting, that's just a little glob of SQL and it'll return a nice list - maybe some HTML to get it formatted pretty into a table or whatever - a lot can really be automated. This is what computers are for, making things easier). I'm not keen on selecting a limited number of questions - grouping many similar to reduce total count and possibly stripping of extra/duplicate detail or rephrasing to suit, sure - but If anyone should, I'd suggest it's Dr Ashurbeyli as he knows how much time he has, and what questions he has capability to provide answers.
If we end up submitting 2000 questions, so be it - it should just be implicit not to expect 2000 answers, just what he can reasonably get through. By not expecting realtime response, and granting a large window(maybe a week after submission, as an optional timeframe more than a rigid requirement, actually released as and when it's done with) in which to respond it should hopefully maximise the number that's possible to reasonably get through. If it's felt important to be answered, it can always be asked again - especially right now there's not a massive rush with most things as they hinge on too many other things being completed first. I'd expect the total number of questions to drop over time to a much more managable level - this sort of thing, IMHO is only a "stopgap measure" - Once we select for ourselves some governmental model then a lot clearer lines can be drawn in a lot of places. If it progresses to the "direct democracy" model then community activity should imply knowlege - or easy access to - of various initatives and projects, progress within and thusly have other places to seek more detailed answers.