So is saving gas the main goal of this EIP? Would it make sense to make such an analysis on real-world code instead of 2-3 hand-crafted examples?
If gas saving is the main goal, then please note that there are still tons of places to optimize existing code without the need to change the EVM.
Furthermore, unless this saves more than 10% gas on real-world examples, I would prefer our workforce to be directed at scaling solutions instead which allow a factor of 10 to 100 in gas savings.