Improve benchmark reporting for contracts

[philippecamacho]

This issue was reported by @alxiong:

also our gas benchmark code is flaky, mostly because forge snapshot is flaky, (the vm.pause/resumeGasMetering() isn't very accurate)

also i find that forge test --gas-report is more accurate than forge snapshot for reasons idk

let's upgrade our forge version soon and use their new vm.lastCallGas() instead: https://github.com/foundry-rs/foundry/discussions/8323#discussioncomment-9927571