Draft: Refactor to use full_blocks and full_transactions to optimize computations

Introduce new full_blocks and full_transaction which combine the signed_block and signed_transaction data structure with the serialized version of those structures. Because we never have to re-serialize the data structures, we can ignore whether they were created using the pre- or post-hf26 serialization. This also lets us compute things like block & transaction ids more efficiently, and cache them and many other things related to transaction validation to avoid repeatedly computing them.