fix(l1): handle missing payloads in engine_getPayloadBodiesByRange

[MegaRedHand]

Motivation

When we added RPC metrics for success/error rates in #5335, we found that we were returning "Internal error" in the engine_getPayloadBodiesByRange method. This was related to changes done in #2430, which turned missing payloads into a DB error. The spec says we should return nulls for any missing payloads, so this deviates from the spec.

Description

This PR fixes the spec deviation by changing the batch-GET methods from the Store API to return None on missing values instead of failing.

Closes #5403

[Copilot]

Pull request overview

This PR fixes a spec deviation in the engine_getPayloadBodiesByRange RPC method by changing batch-GET methods in the Store API to return None for missing payloads instead of returning database errors.

Key changes:

• Modified read_bulk_async to return Vec<Option<V>> instead of failing on missing keys
• Updated get_block_bodies, get_block_bodies_by_hash, and read_fullsync_batch methods across the storage trait and implementations
• Adjusted RPC handler to work with the new signature (removing double-wrapping of Options)

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
crates/storage/api.rs	Updated trait method signatures to return Vec<Option<T>> instead of Vec<T>
crates/storage/store.rs	Updated method signatures to match the trait changes
crates/storage/store_db/in_memory.rs	Implemented the new behavior to return None for missing values instead of skipping them
crates/storage/store_db/rocksdb.rs	Modified read_bulk_async to return None for missing keys and updated get_block_bodies with complex logic to handle missing hashes/bodies
crates/networking/rpc/engine/payload.rs	Simplified the RPC handler by removing the double-wrapping of Options since the storage layer now returns the correct format
crates/networking/p2p/sync.rs	Added error handling to convert None values to MissingFullsyncBatch errors where missing data is unexpected during sync

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[github-actions]

Lines of code report

Total lines added: 22
Total lines removed: 11
Total lines changed: 33

Detailed view

+---------------------------------------------+-------+------+
| File                                        | Lines | Diff |
+---------------------------------------------+-------+------+
| ethrex/crates/networking/p2p/sync.rs        | 1368  | +8   |
+---------------------------------------------+-------+------+
| ethrex/crates/storage/store_db/in_memory.rs | 622   | -11  |
+---------------------------------------------+-------+------+
| ethrex/crates/storage/store_db/rocksdb.rs   | 1725  | +14  |
+---------------------------------------------+-------+------+

[rodrigo-o]

LGTM, this has been running for 3 hours without any internal error. That said we should address Copilot's comment

[Oppen]

Fun fact: I've been working in something related the last few days, though I didn't know if it would be correct to change the read_fullsync_batch part. Can you do the same for the GetBlockBodies RLPx message? With this change it should have the same behavior, and batching the reads should help its performance (we can use batched_multiget, BTW).