feat: stabilize lower_regular_op_cost feature

[matklad]

Feature to stabilize

This PR stabilizes new, lower regular_op_cost. It is lowered from 3_856_371 to 2_207_874. regular_op_cost is roughly the cost of primitive wasm operation, and is a major cost factor for computation-intensive contracts. The underlying change which allowed us to lower the cost is the switch from wasmer0 to wasmer2 implementation, which improved real wasmer perfomance across the board

Testing and QA

Cost measurement was reproduced by several people using both old and new estimator. The cost measurement itself is rather simple (probably the simplest of all our costs), so the relative risk here is low. Additionally, the reduction "makes sense" in terms of various ad-hoc comparisons of wall-clock times I did when comparing wasm runtimes.

Note that we are stabilizing this feature on a very short notice. While this in general is risky, I think it's ok for this particular case, as it is just cost reduction, and there's strong evidence that new wasmer is indeed faster. In a sense, most risk here is actually carried by #4934 , but that was on nightly for a significantly longed time.

Checklist

• No nightly test regression
• Update CHANGELOG.md to include this protocol feature in the Unreleased section.

[matklad]

@bowenwang1996 bad news: I didn't see PR template suggested anywhere, and I had to copy-paste it manually. Good news -- the idea of the template works, going to fill the changelog now as I've forgot to:)

[bowenwang1996]

@bowenwang1996 bad news: I didn't see PR template suggested anywhere, and I had to copy-paste it manually. Good news -- the idea of the template works, going to fill the changelog now as I've forgot to:)

Sad :( Also the checkboxes didn't really work :(

[bowenwang1996]

Looks like I forgot one space...