fix: Race condition while unwinding blocks

[spalladino]

We hit the following error in CI:

19:05:17   ● e2e_block_building › reorgs › detects an upcoming reorg and builds a block for the correct slot
19:05:17 
19:05:17     Could not retrieve body for block 4 0x139a9efee631a725b0ed6bc428460ceedb98bd2502ef13eeaa65123fd0cd98f9
19:05:17 
19:05:17       124 |         const blockBodyBuffer = await this.#blockBodies.getAsync(blockHash);
19:05:17       125 |         if (blockBodyBuffer === undefined) {
19:05:17     > 126 |             throw new Error(`Could not retrieve body for block ${header.globalVariables.blockNumber.toNumber()} ${blockHash}`);
19:05:17           |                   ^
19:05:17       127 |         }
19:05:17       128 |         const body = Body.fromBuffer(blockBodyBuffer);
19:05:17       129 |         const block = new L2Block(archive, header, body);
19:05:17 
19:05:17       at BlockStore.getBlockFromBlockStorage (../../archiver/dest/archiver/kv_archiver_store/block_store.js:126:19)
19:05:17       at BlockStore.getSettledTxReceipt (../../archiver/dest/archiver/kv_archiver_store/block_store.js:165:23)
19:05:17       at AztecNodeService.getTxReceipt (../../aztec-node/dest/aztec-node/server.js:357:34)
19:05:17       at e2e_block_building.test.ts:567:22
19:05:17       at retryUntil (../../foundation/dest/retry/index.js:84:24)
19:05:17       at Object.<anonymous> (e2e_block_building.test.ts:566:7)

Apparently this happens because we try calling getTxReceipt during a block unwind (ie reorg) operation, the new test does not fail if the block is unwound while requesting a tx could reproduce it consistently. This is odd, since the only way that can happen is if the block body has been deleted but not the header. And these operations happen within the same write tx in unwindBlocks.

Either way, this fixes it by ignoring missing block bodies as if the entire block were missing.

[alexghr]

Nice catch!