receipt decoder - handle both formats at read time

[macfarla]

PR description

Fix receipt decode failure after snap sync

Problem

After snap syncing receipts from a peer (eth/69 or eth/70), subsequent reads of those receipts from local storage failed with:

RLPException: Cannot enter a lists, input is fully consumed (at bytes 0-0: [])

This manifested in two ways:

1. Local reads — rlpDecodeTransactionReceipts (called by blockchain.getTxReceipts()) threw RLPException when processing previously snap-synced blocks
2. Serving receipts to peers — EthServer.constructGetPaginatedReceiptsResponse threw RLPException, which bubbled up to EthProtocolManager and triggered a BREACH_OF_PROTOCOL disconnect against the requesting peer , incorrectly blaming them for a local storage problem

Root cause

Snap sync receives receipts in wire format. Compact receipts (eth/69/70, no bloom filter) encode log topics as raw bytes32. The read path used TransactionReceiptDecoder, which calls Log.readFrom(in, compacted=true) and expects canonical compacted storage format where topics are [leadingZeros, shortData] lists. This format mismatch caused the RLPException.

Additionally, for EIP-2718 typed receipts in wire format (type_byte || rlp_list, first byte 0x01–0x7F), TransactionReceiptDecoder reads the first byte as a 1-byte SHORT_ELEMENT, slices it off leaving Bytes.EMPTY, then fails on enterList() of the empty input - the "at bytes 0-0: []" error.

Changes:
Instead of converting sync receipts to canonical format at write time (which fails for wire-format typed receipts passed to TransactionReceiptDecoder), store raw wire bytes and handle both formats at read time:

• putSyncTransactionReceipts: restored to store raw bytes with normalizeReceiptRlpElement wrapping raw type bytes (0x01-0x7F) as byte strings so the RLP list structure remains unambiguous.

• rlpDecodeTransactionReceipts: now uses SyncTransactionReceiptDecoder + SyncTransactionReceiptConverter rather than TransactionReceiptDecoder directly, handling all 8 format combinations (legacy/typed × compacted/non-compacted × wire/canonical).

• SyncTransactionReceiptDecoder.parseLogs: upgraded to handle both raw bytes32 topics (wire format) and [leadingZeros, shortData] lists (canonical compacted storage format) via per-item nextIsList() check and a new readTrimmedData helper mirroring Log.readFrom compacted logic.

• SyncTransactionReceiptConverter.toTransactionReceipt: added storageFormatRlpInput helper to wrap wire-format typed receipts (first byte < 0x80) as byte strings before passing to TransactionReceiptDecoder, fixing the "Cannot enter a list, input is fully consumed (at bytes 0-0: [])" RLPException on fresh nodes.

Fixed Issue(s)

Thanks for sending a pull request! Have you done the following?

• Checked out our contribution guidelines?
• Considered documentation and added the doc-change-required label to this PR if updates are required.
• Considered the changelog and included an update if required.
• For database changes (e.g. KeyValueSegmentIdentifier) considered compatibility and performed forwards and backwards compatibility tests

Locally, you can run these tests to catch failures early:

• spotless: ./gradlew spotlessApply
• unit tests: ./gradlew build
• acceptance tests: ./gradlew acceptanceTest
• integration tests: ./gradlew integrationTest
• reference tests: ./gradlew ethereum:referenceTests:referenceTests
• hive tests: Engine or other RPCs modified?

[Matilda-Clerke]

Isn't it possible to just do this change in the TransactionReceiptDecoder to avoid the need to use SyncTransactionReceiptDecoder and SyncTransactionReceiptConverter when loading from the DB?