abstract over state _and_ storage tries

[0xaatif]

@Nashtare
@hratoanina
I think I've missed something about SMT support.
It looks like SMTs care about all writes to the account's storage:

zk_evm/trace_decoder/src/type2.rs

Lines 158 to 160 in 29aaa1e

	for (slot, value) in storage {
	trie.set(key_storage(addr, slot), value);
	}

I designed trait StateTrie on the assumption that there was a struct that held AccountRlp (or something like it).
My understanding was that storages were out-of-band (BTreeMap<H256, TypedMpt> is sprinkled around the place).
For SMT support, do we care about updating the storage inside the smt?
If so, there will have to be another trait around storage handling, and probably a chunk of design work.

- @0xaatif in Slack

Indeed, you do care about individual storage cells. In the SMT format, the state trie doesn't store an account at a single place, but stores the nonce, the balance, the code hash, the code length and every individual (non-zero) storage cell value at different locations.

- @hratoanina in Slack