[frontier] Add support for PoV gas refunding

[Agusrodri]

What does it do?

Enables PoV gas refunding, which means that the difference between the benchmarked worst-case and the real proof-size consumption is now reclaimed back to the user.

⚠️ Breaking Changes ⚠️

• The main breaking change of this PR consists in the PoV reclaim functionality itself.
• For transactions that consume big PoV (e.g a smart contract calling many other inner smart contracts), and as long as the computed PoV by the host function is lower than the estimated one, users will be refunded with the unused PoV gas.
• Even if the whole functionality of this PR is not expected to break anything, it's worth mentioning it as breaking, as some dapps/projects may notice these changes on their monitoring tools.

Related frontier (fork) PRs

• refund pov gas based on measured proof size usage frontier#230
• Enable ProofSizeExt when estimating gas and emit ReportRefund when PoV gas is the effective gas frontier#235

[github-actions]

WASM runtime size check:

Compared to target branch

Moonbase runtime: 2248 KB (no changes) ✅

Moonbeam runtime: 2224 KB (no changes) ✅

Moonriver runtime: 2224 KB (no changes) ✅

Compared to latest release (runtime-3300)

Moonbase runtime: 2248 KB (+220 KB compared to latest release) ⚠️

Moonbeam runtime: 2224 KB (+228 KB compared to latest release) ⚠️

Moonriver runtime: 2224 KB (+232 KB compared to latest release) ⚠️

[github-actions]

Coverage Report

@@                       Coverage Diff                        @@
##           master   agustin-frontier-pov-reclaim      +/-   ##
================================================================
+ Coverage   74.75%                         78.98%   +4.23%     
- Files         369                            303      -66     
- Lines       94246                          88231    -6015     
================================================================
- Hits        70453                          69685     -768     
- Misses      23793                          18546    -5247     

Files Changed	Coverage
/node/cli/src/cli.rs	33.33% (+0.40%)	🔼
/node/cli/src/command.rs	17.61% (+0.71%)	🔼
/node/service/src/lib.rs	62.25% (+0.23%)	🔼
/precompiles/relay-data-verifier/src/mock.rs	87.69% (+2.16%)	🔼
/runtime/moonbeam/src/xcm_config.rs	45.45% (+2.18%)	🔼
/runtime/moonbeam/tests/common/mod.rs	94.68% (-0.06%)	🔽
/runtime/moonriver/src/xcm_config.rs	45.45% (+2.18%)	🔼
/runtime/moonriver/tests/common/mod.rs	94.80% (-0.05%)	🔽

Coverage generated Wed Nov 27 14:51:19 UTC 2024

[crystalin]

I tested a live transaction (with lazy fork) from BeamSwap using this PR.
The cost goes from 0.48 GLMR to 0.041 GLMR !!!
(current runtime LEFT), (this PR RIGHT)

[crystalin]

@Agusrodri I can see the ref_time of weight (v2) jumps from 93M to 7M. Is that normal ?

[Agusrodri]

@Agusrodri I can see the ref_time of weight (v2) jumps from 93M to 7M. Is that normal ?

Not sure if this is normal to be honest. From my understanding the refund should only take place in the pov size right? Will need to investigate.

[crystalin]

After more consideration, it is expected to also be lower. For Ethereum transaction, the weight (ref_time) is computed from the gas cost, using a gas to weight ratio. The gas being lower (due to PoV) the ref_time is lower being lower. So all good