refactor Metadata._get_role_and_keys()

[jku]

EDIT: This relates to code in #2165 -- should likely be fixed in that PR

Looking at Metadata._get_role_and_keys() after a while... I did not do a good job of it.

Potential improvement:

• it should return only the keys for the delegated role (instead of all delegated keys and the specific Role). I was probably avoiding unnecessary allocations here but this makes the code harder to read
• does not have to return the whole role but does have to return threshold instead
• the method should be public as it's useful in applications as well
• these changes mean some changes in verify_delegate()

Something like this

    def get_keys_and_threshold(self, delegated_role: str) -> Tuple[List[Key], int]:

[lukpueh]

FYI, I also changed this method in #2246 to get keys and threshold from an explicitly passed Signed object instead of self.signed.

I needed this for a common abstraction of Metadata and Envelope (dsse) classes, which accesses self.signed via getter method, that, in the case of Envelope, takes an optional (deserialization) argument.

[jku]

That leads to the conclusion that metadata is the wrong location to begin with? I think I might agree with that: Targets and Root objects could both implement a method with the same signature (in effect implementing a hypothetical Delegator Protocol)

[jku]

I just tried this and it looks a lot simpler: will fix this in the no-key branch/PR

[jku]

closing: this is reviewed as part of #2165