The performance cores got a bit faster with the clock increase and performance per watt decreased a bit. But they made up for it with the improved efficiency cores that actually do improve performance/watt. There is no point to introducing a new SoC that doesn't have improvements in performance. They gained in GPU as well. I suppose they could have left the performance cores at M1 levels and just banked on the improved efficiency and GPU cores but they must have decided that they had enough cooling to handle a bit more CPU as well. Why does it have to scale linearly to be useful?
I think that they had planned on being on 3 nm for the M2 and the better process would have offset the higher frequency. Oh well.