endorser/execution: cover DualStateDB to 98% with dual-side assertions

[Storm1289]

Summary

Closes #320.

Adds endorser/execution/dual_statedb_test.go. The tests build both a go-ethereum in-memory state.StateDB and the endorser SnapshotDB, wrap them in a DualStateDB, and assert that every mutation is reflected consistently on both sides.

Also adds four focused tests for EVMEngine.BalanceAt, StorageAt, CodeAt, and NonceAt, covering the fresh-snapshot read path.

Coverage delta

File	Before	After
dual_statedb.go (issue's primary target)	0%	98.3%
statedb.go	~78%	83.3%
Package total	22.3%	47.4%

What's tested

All method categories from the issue are covered:

• Accounts: CreateAccount, CreateContract, Exist, Empty, Touch
• Balances/nonces: AddBalance, SubBalance, SetNonce
• Code: SetCode, GetCodeSize, GetCodeHash
• Storage: SetState, GetStateAndCommittedState, GetStorageRoot, transient state
• Refunds
• Self-destruct
• Access list, including Prepare precompile seeding
• Snapshot / revert round-trip
• Logs
• Finalise
• Result
• Logs
• IsNewContract
• LogsForBurnAccounts
• Witness
• AccessEvents

Open question for the maintainer

The package-level coverage is capped by the two *_logger.go files (eth_statedb_logger.go and statedb_logger.go), which the issue explicitly lists as out of scope. Together they are ~770 LOC of pure logging pass-throughs (~22% of the package by size), gated behind the DebugLogs config, and both remain at ~0%.

Two ways to reconcile the 70% target while keeping those files out of scope:

1. Add endorser/execution/*_logger.go to the codecov.yml ignore list, matching the existing **/*.pb.go / **/*.sql.go pattern for non-review-worthy code. In-scope coverage would then be represented accurately.
2. Interpret the 70% target as applying to the in-scope files, where coverage is already ~90%.

I didn't touch codecov.yml in this PR. Happy to add the ignore lines in a follow-up commit if you prefer option 1.

Out of scope

• eth_statedb_logger.go / statedb_logger.go - debug wrappers, config-gated
• Integration-level EVM execution - covered by the integration suite

Test plan

• go build ./...
• go test -race -count=1 ./endorser/execution/...
• go test -cover ./endorser/execution/... → 47.4% package, 98.3% on dual_statedb.go
• CI Quick Tests

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 36.62%. Comparing base (09ef2a0) to head (eaf056c).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #345      +/-   ##
==========================================
+ Coverage   34.07%   36.62%   +2.54%     
==========================================
  Files          82       82              
  Lines       11652    11652              
==========================================
+ Hits         3970     4267     +297     
+ Misses       7347     7042     -305     
- Partials      335      343       +8     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[github-actions]

🎉 OZ Hardhat Compatibility — 11 tests fixed!

Stale baseline entries (11) — remove these

• AccessManager admin operations subject to a delay #setRoleAdmin restrictions when the call does not come from the manager (msg.sender != manager) when the function requires the caller to be granted with a role other than PUBLIC_ROLE when the required role is granted to the caller when role granting is not delayed when caller has no execution delay succeeds via execute
• AccessManager admin operations subject to a delay #setRoleGuardian restrictions when the call does not come from the manager (msg.sender != manager) when the function requires the caller to be granted with a role other than PUBLIC_ROLE when the required role is granted to the caller when role granting is not delayed when caller has no execution delay succeeds via execute
• AccessManager admin operations subject to a delay #setGrantDelay restrictions when the call does not come from the manager (msg.sender != manager) when the function requires the caller to be granted with a role other than PUBLIC_ROLE when the required role is granted to the caller when role granting is not delayed when caller has no execution delay succeeds via execute
• AccessManager admin operations subject to a delay #setTargetAdminDelay restrictions when the call does not come from the manager (msg.sender != manager) when the function requires the caller to be granted with a role other than PUBLIC_ROLE when the required role is granted to the caller when role granting is not delayed when caller has no execution delay succeeds via execute
• AccessManager admin operations not subject to a delay #setTargetFunctionRole restrictions when the call does not come from the manager (msg.sender != manager) when the function requires the caller to be granted with a role other than PUBLIC_ROLE when the required role is granted to the caller when role granting is not delayed when caller has no execution delay succeeds via execute
• AccessManager admin operations not subject to a delay role admin operations #grantRole restrictions when the call does not come from the manager (msg.sender != manager) when the function requires the caller to be granted with a role other than PUBLIC_ROLE when the required role is granted to the caller when role granting is not delayed when caller has no execution delay succeeds via execute
• UUPSUpgradeable upgrade to upgradeable implementation with call
• ERC721Wrapper onERC721Received mints a token to from
• Create2 deploy deploys a contract without constructor
• GovernorProposalGuardian using $ERC20Votes "before each" hook for "deployment check"
• GovernorProposalGuardian using $ERC20VotesTimestampMock "before each" hook for "deployment check"

Full breakdown — 5519 passed, 874 failed, 1 skipped (6394 total, 86.3% passing)

By suite

• access: 408/603 passing (68%)
• crosschain: 44/48 passing (92%)
• finance: 8/16 passing (50%)
• governance: 243/532 passing (46%)
• metatx: 26/32 passing (81%)
• proxy: 287/320 passing (90%)
• token: 1800/1882 passing (96%)
• utils: 2703/2960 passing (91%)

Expected failures by cause (608)

• max-code-size: 212
• execution reverted: 129
• hardhat-time-rpc: 93
• fixed-timestamp: 35
• subcall-effect-lost: 31
• fixed-block-number: 23
• hardhat_setStorageAt: 17
• gas required exceeds allowance (10000000): 12
• eth_getProof: 6
• no private key available for address 0x8B148f5D02D560f1f46438656a85C2d3cC143390: 5
• Expected transaction to be reverted with custom error 'FailedCall', but it reverted with custom error 'TimelockUnexpectedOperationState': 4
• Expected transaction to be reverted with custom error 'InvalidAccountNonce', but it reverted with custom error 'ERC3009InvalidAuthorizationTime': 4
• evm_setAutomine: 3
• execfailure-dropped-not-committed: 3
• no private key available for address 0xB40Dc4D4FE972F7C3BD3979d0B5E9582073E6E7E: 3
• out of gas: 3
• transaction already pending: 3
• Expected transaction to be reverted with custom error 'FailedCall', but it didn't revert: 2
• estimate-gas-allowance-not-revert: 2
• hardhat-block-rpc: 2
• hardhat_impersonateAccount: 2
• hardhat_setBalance: 2
• no private key available for address 0x09120eAED8e4cD86D85a616680151DAA653880F2: 2
• no private key available for address 0x0B306BF915C4d645ff596e518fAf3F9669b97016: 2
• Expected transaction to be reverted with custom error 'ERC3009InvalidAuthorizationTime', but it didn't revert: 1
• Expected transaction to be reverted with custom error 'TimelockUnexecutedPredecessor', but it reverted with custom error 'TimelockUnexpectedOperationState': 1
• Expected transaction to be reverted with panic code 0x01 (Assertion error), but it reverted with a custom error: 1
• could not decode result data (value="0x", info={ "method": "getValue", "signature": "getValue()" }, code=BAD_DATA, version=6.16.0): 1
• expected '0xaeee8799c04cdd8ea4d908c7d22d99ea63c…' to equal '0x7f6f6e29c15580dc65a7ad00b0ca049e17a…': 1
• expected 36000 to equal 4919.: 1
• expected 4919 to equal 0.: 1
• no private key available for address 0xc6e7DF5E7b4f2A278906862b61205850344D4e7d: 1

Lock in the win — run go run ./cmd/baseline update --suite oz-hardhat and commit the updated baseline to remove these.

[github-actions]

Perf Smoke — CI baseline (lower-bound against test committer)

Metric	Value
Throughput	520.90 tx/s
Invalid rate	0.000000
Conflict rate	0.000000

_{Workflow run}

[arner]

Nice! Also functions as documentation on which features we do implement and what we skip.

[arner]

LGTM!