In our original scorer_identifier, I explicitly removed pyrit_version from the identifier because we don't want it as part of the hash (e.g. if the pyrit_version changes, we don't want to have to re-calculate scorer metrics).

Now that we have more robust general Identifiers, this PR adds it back.

Also added tests and refactored for readability (clearly separating hashes and storage metadata exclusion keys).

My more overarching question is when would pyrit_version be used in practice? I figured that our identifiers have enough granular information that a broader pyrit_version wouldn't be needed but I might be missing something.