If M2 is based on A15 (which I don't think it will be) and subsequently on TSMC N5, then we already know from iPhone benchmarks the advantage will be negligible (<5%), especially for the CPU side. If, however (what I believe), M2 is based on A16 and with that (most likely) TSMC N3, then you can expect a 15% uplift in performance from the process alone, plus probably around 5% to maybe, possibly 10% from incremental, architectural improvements considering A16 is two iterations up from A14. So I fully expect M2 to be around 20% better in terms of single thread performance. Things look even better on the GPU side as the A15 GPU cores are already quite a bit stronger than A14's.
So.... if you really want a thin and light Apple Silicon device you should probably wait for september/october (if you can), unless you really dig the current MBA design and/or plan on mostly using it as a lightweight task machine (browsing, office, web development, etc). Plus: I cannot imagine the MacBook Air successor (I suspect just MacBook) to come as cheap as the current M1 MBA. Expecting a 200 bucks price bump seems reasonable given the current "state" of the world, the fact that TSMCs N3 process is significantly more expensive, and that we can expect M2 to utilize more expensive LPDDR5 memory (opposed to current M1 devices using LPDDR4x).
It also makes sense that they'd offer the MBA (or just MB) with M2 first since it is by far the best selling Mac out there, and the only one to garner substantial market share. Personally I think we will not see another Mac Mini non-pro, as the desktop market is firmly in the hand of people using their machines for business and professional work, and hence will be willing to pay for a pro level machine. I guess the M1 Mac Mini was just a one-shot stopgap solution to bring "some" Apple Silicon to desktop users right away.
2 cents.