Improve how declare handles an existing class hash

[ericglau]

When I call declare on an implementation contract to be used with a proxy, if it was previously declared, register_class_hash currently raises an exception containing the hash in the message string.

To improve usability, I want a way to get the existing declaration's class hash (without having to parse the error message) so that it can be used afterwards.

Consider one of the following:
1. Use a custom error that contains the hash in a field, so that the caller of declare can catch the error and retrieve the hash
2. In declare, suppress the error if a class hash already exists, and just proceed to return with the existing class hash
3. In declare, suppress the error if a class hash already exists, and update the entry in the declaration file (e.g. if a different alias is provided)
4. In declare, have an optional flag such as update which allows it to do (3) only if the flag is set.
5. Provide a function similar to get_declaration, but takes a contract (rather than an alias) and returns the corresponding class hash.

[ericglau]

I prefer option (4) above, since an update flag would make declare easier to use in this scenario, and still retain the original error if an update is not desired.

[martriay]

the equivalent of (5) would be:

from starkware.crypto.signature.fast_pedersen_hash import pedersen_hash
from starkware.starknet.core.os.class_hash import compute_class_hash

def get_hash(contract_class):
    return compute_class_hash(
        contract_class=contract_class,
        hash_func=pedersen_hash
    )