feat: update representation-independent hash computation

[mraszyk]

This MR updates the specification of the representation-independent hash computation for the case of maps with nested maps.

[adamspofford-dfinity]

Should it not be the other way around, H(key) . hash_of_map(...)? It seems inconsistent that arrays are hashed once but maps are hashed twice. (Plus if there's a double hash it can no longer be implemented as a single stateless serde serializer.)

[mraszyk]

Should it not be the other way around, H(key) . hash_of_map(...)? It seems inconsistent that arrays are hashed once but maps are hashed twice. (Plus if there's a double hash it can no longer be implemented as a single stateless serde serializer.)

Sounds good! I've updated the PR and its example computation to reflect your suggestion.

[Dfinity-Bjoern]

As discussed, looks good, thanks @mraszyk!