New payload with witness ssz

[hudem1]

Follow up of PR #11623 as this was from a forked repo

What types of changes does your code introduce?

• Bugfix (a non-breaking change that fixes an issue)
• New feature (a non-breaking change that adds functionality)
• Breaking change (a change that causes existing functionality not to work as expected)
• Optimization
• Refactoring
• Documentation update
• Build-related changes
• Other: Description

Testing

Requires testing

• Yes
• No

If yes, did you write tests?

• Yes
• No

Notes on testing

Optional. Remove if not applicable.

Documentation

Requires documentation update

• Yes
• No

If yes, link the PR to the docs update or the issue with the details labeled docs. Remove if not applicable.

Requires explanation in Release Notes

• Yes
• No

If yes, fill in the details here. Remove if not applicable.

Remarks

Optional. Remove if not applicable.

[LukaszRozmej]

Please trim comments and remove the parts that are self-explanatory in code or unnecessary

[LukaszRozmej]

Maybe merge after #12193 and updating to newest schema.

in test prefer TestCaseData over object[]

[LukaszRozmej]

@claude re-review

[claude]

Claude finished @LukaszRozmej's task in 6m 46s —— View job

---

Re-review — engine_newPayloadWithWitness (SSZ)

Re-reviewed the current state against origin/master. The architecture has been substantially reworked since the first two review rounds and the prior blocking findings are resolved. No Critical or High issues remain.

Status of prior findings

ID	Finding	Status
H1	DI lambda anti-pattern in BlockProcessingModule	✅ Resolved — main path is back to .AddScoped<IBlockAccessListManager, BlockAccessListManager>(); the code-repo choice is now a CodeInfoRepositoryFactory delegate (Caching default).
H2	fallbackTrieStore never disposed	✅ Resolved — WitnessCapturingBlockProcessingEnv.Graph.Dispose and WitnessGeneratingBlockProcessingEnvFactory.PooledEntry.Dispose both dispose the trie store.
M1	LINQ on hot path	✅ Resolved — recorder deleted; foreach over plain dictionaries.
M2	Fragile (WitnessCapturingWorldStateProxy) cast	✅ Resolved — new design uses a plain AddDecorator<IBlockProcessor, WitnessCapturingBlockProcessor>() + a separate root-scope witness graph; no cast.
M3	Fragile is NewPayloadWithWitnessSszHandler route exclusion	✅ Resolved — special case gone; routing is fully abstracted behind ISszEndpointHandler (HttpMethod/Resource/Version/AcceptsPathExtra). Handler is now generic NewPayloadWithWitnessSszHandler<TVersion,TWire>.
M4	TryArm memory-ordering comment	✅ Resolved — WitnessRendezvous rewritten as a lock-guarded single hash-validated slot; no CAS, self-cleaning WitnessRequest/using.
L2	node.Keccak null guard	✅ Resolved — file deleted.
L3	No timeout before cancel	✅ Resolved — replaced by using WitnessRequest + non-blocking IsCompletedSuccessfully probe; documented.

Also verified: the witnessMode/_isBuilding parallel gate was replaced with the precise _hasParentReaderPool precondition (per @LukaszRozmej's review), and the witness Witness is disposed on both transports — SSZ via using (result) in WriteSszResultAsync, JSON-RPC via ResultWrapper.DisposePayloads.

Remaining (all Low / non-blocking)

• DI lambda in the two witness env builders (WitnessCapturingBlockProcessingEnv.cs:87, WitnessGeneratingBlockProcessingEnvFactory.cs:94) — same hand-listed-args pattern as H1, now confined to the specialized witness scopes. Cleanly replaceable by registering CodeInfoRepositoryFactories.Witness in the scope and reverting to the one-liner. See inline comment.
• validation_error capped at 1024 (SszValidationError = SszList(1024), matching the truncation) — internally consistent and matches the other endpoints, but execution-apis#773 discussion suggests 8192. Already tracked by @LukaszRozmej ("update to newest schema" before merge); no code bug.
• engine_newPayloadWithWitness hardcodes engine_newPayloadV5 — acknowledged by the author as acceptable until a V6 witness variant exists.
• Tests — prefer TestCaseData over object[] and dedup where noted; style only, already acknowledged.

Correctness spot-checks (no issues found)

• Single execution: witnessed block runs only through witness.Processor.ProcessOne; inner is skipped. Recorder wraps the real main world state, so it is imported, not re-executed.
• PatriciaTrieWitnessGenerator mirrors BulkSet, handles inserts (created accounts/slots → Read), deletes (→ collapse-sibling capture), and the empty-touch root anchor; production walks use parallelize: false with a non-thread-safe sink, as documented. Backed by dedicated tests.
• Read/Delete classification via post-state AccountExists/Get bypasses the recording overrides (base.*), so it doesn't pollute the accessed set.
• Rendezvous: block processing is serialized, a second concurrent request is cleanly declined (VALID, no witness), and every exit path releases the slot via the using registration.

Verdict: Mergeable — no Critical/High/Medium findings outstanding. Suggest folding in the small DI-lambda cleanup and confirming the final validation_error limit against the latest schema before merge, per @LukaszRozmej.