Minimal decoding of receipts during syncing

[jframe]

During the initial sync, the decoding and reencoding of receipts use a lot of memory which limits the parallelization that can be used during syncing. If we are to decode only the data enough to verify the receipt and insert the already encoded receipt RLP into the database without reencoding then this could reduce the amount of memory and time needed to decode and reencode during syncing.