feat(rpc): implement Filecoin.StateMarketStorageDeal

[hanabi1224]

Summary of changes

Blocked on ChainSafe/fil-actor-states#267

Changes introduced in this pull request:

• Implement V0 RPC method Filecoin.StateMarketStorageDeal
• Api compare tests

Test output (with fil-actor-states on ChainSafe/fil-actor-states#267 branch)

➜  forest git:(hm/rpc-state-market-storage-deal) ✗ cargo run --bin forest-tool -- api compare --filter StateMarket ~/fr/
snapshots/calibnet/forest_snapshot_calibnet_2024-03-19_height_1450304.forest.car.zst
    Finished dev [unoptimized] target(s) in 0.39s
     Running `target/debug/forest-tool api compare --filter StateMarket /home/me/fr/snapshots/calibnet/forest_snapshot_calibnet_2024-03-19_height_1450304.forest.car.zst`
| RPC Method                            | Forest | Lotus |
|---------------------------------------|--------|-------|
| Filecoin.StateMarketStorageDeal (100) | Valid  | Valid |

Reference issue to close (if applicable)

Closes

Other information and links

Change checklist

• I have performed a self-review of my own code,
• I have made corresponding changes to the documentation,
• I have added tests that prove my fix is effective or that my feature works (if possible),
• I have made sure the CHANGELOG is up-to-date. All user-facing changes should be reflected in this document.

[LesnyRumcajs]

# Integration tests / Offline RPC check (pull_request) Failing after 2m
+ forest-tool api compare ./forest_snapshot_calibnet_2024-03-19_height_1450337.forest.car.zst --forest /ip4/127.0.0.1/tcp/8080/http --lotus /ip4/127.0.0.1/tcp/8081/http --n-tipsets 5
Error: unimplemented

no green checkmark, no review!

[hanabi1224]

no green checkmark, no review!

@LesnyRumcajs Yeah, need to incorporate ChainSafe/fil-actor-states#268 to make it green. unimplemented is from here

// `get_proposal_array` does not exist for V13
State::V13(_st) => anyhow::bail!("unimplemented"),

[hanabi1224]

no green checkmark, no review!

@LesnyRumcajs all green now!

[LesnyRumcajs]

I'm not sure about this. Tests need to be deterministic if possible. Otherwise, we may end up with flaky tests that are not reproducible.

If we need shuffling, using a pre-determined seed for the RNG might be a way out. What do you think?

[hanabi1224]

In practice, the snapshot we use (and the tipset to test against) is rolling as well, it's not likely possible to really fix the test cases

[LesnyRumcajs]

But given the same snapshot (e.g., in a local test), making the seed constant would make the test reproducible, right? Or would it still be completely random from network conditions?

[hanabi1224]

But given the same snapshot (e.g., in a local test), making the seed constant would make the test reproducible, right?

@LesnyRumcajs That's true, I can remove the shuffle to make it reproducible for a fixed snapshot.