fix!: Revise VM storage interface

[slowli]

What ❔

Replaces storage slot data returned by StorageInterface with a dedicated type (essentially, (U256, bool)).

Why ❔

Currently, StorageInterface reads storage slots as Option<U256> where None corresponds to initial writes. This loses information during sandboxed execution. Indeed, there may be slots which have non-zero value, but still are initial writes (i.e., these slots were first written to during previous blocks / transactions in the sandboxed batch).

Checklist

• PR title corresponds to the body of PR (we generate changelog entries from PRs).
• Tests for the changes have been added / updated.
• Documentation comments have been added / updated.
• Code has been formatted via cargo fmt and cargo clippy.

[joonazan]

For maximum efficiency with Postgres there'd have to have one method returning (U256, bool) and another returning just U256. I was planning to do that optimization at some point but it is quite complex, so I think it might not be worth it.

I would write the integration of this so that RocksDB/hashmap cache is only queried once per method call. That would help performance a lot, at least according some benchmark flamegraphs.

[slowli]

For maximum efficiency with Postgres there'd have to have one method returning (U256, bool) and another returning just U256.

That's what I've implemented 🙃 I agree that it looks like the most efficient approach. Regarding changes in the Era codebase, that's a separate topic that we can approach iteratively.