Its kinda hard to find something to do that NO ONE else has ever done before, but its very easy to make huge improvements to their systems. And I agree, using bacteria to solve a problem is a very very good idea, its something that people can relate to, which is HUGELY important in a science fair. If making something for a third world country, I'd probably recommend trying out a system in Bacillus. It can sporulate easily making distribution very cheap. Perhaps just take a detection system used in E coli and transfer it to Bacillus, and give the pitch that now in Bacillus it can actually be used in these countries. (But ya, I am a little biased towards Bacillus)
It's easier said then done because of these huge pathways
On experiment 3, I think I heard an iGem project that was thinking about that. The biggest problem is getting that through clinical trials. It might cause a lot of allergic reactions, which could halt it.
Keep in mind to keep experiments almost never go as planned, as I've found out at an academic lab. Try to reduce steps in any experiment, make it simple, and check absolutely everything possible with sequencing and other analytical techniques. It once took me nearly a month to clone a set of only 8 plasmids. Don't get discouraged though: Synthetic biology and science fairs have been the most rewarding experience I've ever had. Every single experiment is worth it.
-Koeng