feat(generator): delegate REST transcoding to google-api-core

[hebaalazzeh]

this PR refactors generator templates to delegate REST transcoding to the centralized helper in google-api-core.

Scope of Changes:

• _shared_macros.j2: Replaced inline static helper invocation with gapic_v1.rest_transcoding.transcode().
• rest_base.py.j2: Stripped the redundant _get_transcoded_request, _get_request_body_json, and _get_query_params_json helper functions.
• _rest_mixins_base.py.j2: Stripped the redundant static helpers from API REST mixins base.
• rest.py.j2 & rest_asyncio.py.j2: Passed the configuration flags (like opts.rest_numeric_enums) into the transcoding macro.
• noxfile.py: Configured local packages/google-api-core to be installed during virtualenv test generation so the new transcoding runtime module is available.

related PR: #17765

[gemini-code-assist]

Code Review

This pull request refactors REST transcoding in the GAPIC generator by centralizing the transcoding logic into a new transcode function within google-api-core (gapic_v1.rest_transcoding). This change eliminates redundant static helper methods (_get_transcoded_request, _get_request_body_json, and _get_query_params_json) from the generated transport templates. Additionally, the PR updates test configurations and introduces comprehensive unit tests for the new transcoding utility. The feedback suggests optimizing the transcode function by using json_format.MessageToDict directly instead of the inefficient round-trip serialization of json.loads(json_format.MessageToJson(...)).

[gemini-code-assist]

Using json.loads(json_format.MessageToJson(...)) to convert a protobuf message to a dictionary introduces unnecessary serialization and deserialization overhead. We can use json_format.MessageToDict directly, which is much more efficient and cleaner.

        query_params_json = json_format.MessageToDict(
            transcoded_request["query_params"],
            use_integers_for_enums=rest_numeric_enums,
        )