chore: Remove Optimism_SpokePool & related contracts

[pxrl]

Optimism_SpokePool was required as an extension of Ovm_SpokePool in order to support Optimism-specific bridges for early ERC20 tokens - notably DAI and SNX. Both of these tokens have been deprecated by Across and the SpokePool no longer holds any balances for them. Reverting to the OP_SpokePool contract is a simplification in terms of code surface area and deployment overhead.

[pxrl]

@droplet-rl Verify that there are no storage layout changes by updating the Optimism SpokePool to an OP_SpokePool implementation. Be absolutely critical and justify your conclusion.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7cd339e9df

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[droplet-rl]

Verified: no storage layout changes. Optimism_SpokePool and OP_SpokePool produce byte-identical compiler-generated storage layouts. Evidence, from weakest to strongest:

1. Checked-in snapshots match. The deleted storage-layouts/Optimism_SpokePool.json vs the existing storage-layouts/OP_SpokePool.json: identical except the contract-name path in each entry's contract field.

2. Fresh compilation at the merge base (fed107d0), where both contracts coexist. I ran forge inspect <Contract> storageLayout --json (solc 0.8.30 per foundry.toml) for both and diffed the full output — including the types section and struct member encodings that checkStorageLayout.sh strips via jq del(.type)/del(.types). This matters: the CI check alone could not catch a same-slot/same-label type change, so I deliberately compared the unstripped layouts. After normalizing only astId (compiler-run noise) and the contract-name string, the layouts are identical across all 33 entries: every slot, offset, label, type encoding, and struct member matches, ending at the inherited __gap at slot 3166.

3. Compilation at this PR's HEAD. OP_SpokePool compiled at 7cd339e9 is identical to Optimism_SpokePool compiled at base — i.e., nothing this PR removes affects the surviving contract's layout. The only non-deletion source change in contracts/ is a comment edit in chain-adapters/CrossDomainEnabled.sol, which is not in the spoke pool inheritance chain.

Why this is structurally expected: neither contract declares any storage of its own. Optimism_SpokePool differed from OP_SpokePool only by two constants (SYNTHETIX_BRIDGE, SNX — constants are inlined, never stored) and a function override (code, not storage). Both inherit their entire layout from the same Ovm_SpokePool → SpokePool + CircleCCTPAdapter chain, compiled from the same sources, and both call __OvmSpokePool_init(..., Lib_PredeployAddresses.OVM_ETH).

What layout equality does not cover — three caveats for the upgrade itself:

• Immutables. wrappedNativeToken, the quote/fill buffers, USDC, and the CCTP token messenger are immutables in implementation bytecode, not proxy storage. The OP_SpokePool implementation deployed for chain 10 must be constructed with Optimism's values. DeployOPSpokePool.s.sol sources these per-chain, so a chain-10 deploy gets the right ones — but this is a deploy-parameter check, not something layout equality proves.
• Deployed-impl assumption. My comparison proves repo-code equivalence. It equals the on-chain implementation's layout only via the CI snapshot gate (.github/workflows/pr.yml:211), which has enforced storage-layouts/Optimism_SpokePool.json on every PR — both snapshot files have moved in lock-step through the same commits (improve: rename Spoke Pool's hubPool variable to withdrawalRecipient #629, feat: EVM changes required to support Solana #672, feat: Add EVM Universal Adapter and remove deposit whitelist checks #974, feat: [POST AUDIT] OFT changes #1062) since the check was introduced in improve(forge-ci): Ignore astId and other non-storage slot related props when diffing storage layouts #523. Any drift there would be a pre-existing issue, not one introduced here.
• One behavioral (non-storage) delta. Optimism_SpokePool overrode _bridgeTokensToHubPool to route SNX (0x8700...99B4) through the Synthetix bridge. OP_SpokePool drops this, so any SNX repayment would attempt the standard OVM bridge path. If the SNX route is fully deprecated this is moot, but worth a one-line confirmation before upgrading.

initialize() is textually identical in both, and moot regardless — the proxy is already initialized and the initializer modifier blocks re-execution.

Conclusion: upgrading the Optimism SpokePool proxy to an OP_SpokePool implementation is storage-layout-safe. No code changes needed on this PR for that purpose.

[droplet-rl]

🔎 View trace