Update selector computation

[LHerskind]

The current function selector computation is not ideal.
It is weird that it is using field while the noir code is using Field and if using a struct, the function signature which is used for the selector just inserts struct meaning that two functions with the same name that have structs as input will have same selector even if different structs.

Function Signature: transfer(struct,struct,field)
Function Selector:  0x6f9c917c

Similar if using arrays. In the function below, every param is different types, but you don't know from the signature or selector

#[aztec(private)]
fn test_selector(
    _struct: AztecAddress,
    _other_struct: EthereumAddress,
    _field: Field,
    _u8: u8,
    _array: [Field; 2],
    _u8_array: [u8; 2],
) {}

Function Signature: test_selector(struct,struct,field,integer,array,array)
Function Selector:  0x1b5ca441

The function signature should be unambiguous and look closer to the noir code that is represents.

---

Proposed solution

I propose that we add struct as tuples, that integers also have their width, and that arrays are inserted similarly to their definitions with sizes. An example is seen below:

struct Asset {
    interest_accumulator: u120,
    last_updated_ts: u120,
    loan_to_value: u120,
    oracle_address: Field,
}

struct Position {
    collateral: Field,
    static_debt: Field,
    debt: Field,
}

struct Wrapped {
    asset: Asset,
    position: Position,
    array: [Field; 2],
}

#[aztec(private)]
fn abi_tester(
    asset: Asset,
    wrapped: Wrapped,
    owner: Field,
    small: u8,
    array: [Field; 2],
    u8array: [u8; 2],
    u120array: [u120; 2],
) {}

// Current
// Function Signature: abi_tester(struct,struct,field,integer,array,array,array)
// Function Selector:  0xd1b771b8

// My proposal
// Function Signature: abi_tester((u120,u120,u120,Field),((u120,u120,u120,Field),(Field,Field,Field),[Field;2]),Field,u8,[Field;2],[u8;2],[u120;2])
// Function Selector:  0xaec244be