I remember helping a friend in med school study for the pharmaceutical stuff by making mnemonic devices for all the root words for various drugs, and turning them into creatures. They were pretty terrible and probably not helpful (I remember LobStar, but I don't remember what he was supposed to help us remember!) All the same, the idea of turning pharmaceuticals into mnemonic monsters always stuck with me, so your pokemon analogy sounds delightful.
Specifics:
1. What kind of game might accomplish my goal of commiting to memory many facts?
I think your idea of turning memorizable facts into stats is a good start. The challenge you may face (and the challenge of many educational games), is these attributes may not directly translate into a well-balanced game experience. But like I said, educational games deal with this all the time, so it might be worth it to check out something like theIGDA Learning and Education Games SIG, or a similar group, to address that angle of it
2. Would a game similar to Pokemon be buildable with Construct 2? What are the strengths and limitations of this program building such game? Any tutorials?
I think Construct 2 would be fine for this, because the logic is very turn-based, so you wouldn't need super high performance. No tutorials come to mind offhand though.
3. How well would this sort of game translate to touch screen platforms?
If it's very menu-based, I don't think it would be a problem at all. For the "moving around the world" part you could experiment with more touch centric controls (touch a place and the player moves to that spot, for example), but you could go either way.