Fix tuple encoding in precompiles return position

[nanocryk]

What does it do?

In Solidity tuple encoding depends on if the content has static size or not. Dynamic size tuples are encoded with an offset to the tuple data, while static size tuples the fields are encoded inlined.

Function returning multiple values look like they are returning tuples, however they are always encoded inline regardless of if the size is dynamic or not. Note that while structs are usually encoded like tuples, using a struct as return will be encoded differently if the struct is dynamic sized or not.

This PR makes the following changes:

• Add a new function encode_as_function_return_value that performs the correct encoding. It relies on a new function in EvmData trait called is_explicit_tuple that should only return true for tuples (it has a default implementation to false so that implementing the trait on other types don't require to think about it). In the case that the data being provided to encode_as_function_return_value returns true for is_explicit_tuple and false to has_static_size, the first 32 bytes (the offset) is stripped.
• This issue was noticed with a function of the Referenda precompile, which returns a dynamic-sized tuple. To keep the same ABI, its return type is changed to an equivalent struct.

What important points reviewers should know?

Is there something left for follow-up PRs?

• EvmData derive macro

What alternative implementations were considered?

Are there relevant PRs or issues in other repositories (Substrate, Polkadot, Frontier, Cumulus)?

What value does it bring to the blockchain users?

[notlesh]

Could this be done as a derive macro?

[nanocryk]

We can add an EvmData derive macro in a follow up PR yes :)