fix(auth): support modern async token responses - #18554

Using google.auth.aio.transport.aiohttp.Request for an async token grant currently raises TypeError because the token helper awaits the response's streaming content() generator and expects the legacy status attribute. Read modern responses through read() and status_code, while retaining the legacy content()/status path and its compressed-body decoding.

The regression tests exercise successful, text, JSON, error and retry responses through both interfaces, the concrete modern aiohttp response adapter, and legacy gzip decoding. A localhost check using both real request adapters succeeds for plain and gzip token responses after the change; the modern adapter failed for both before it.

Validation (Python 3.12):

• Before the fix, the selected regression tests produced 7 failures for modern responses and 7 passes for legacy responses.
• 186 async OAuth and relevant modern/legacy transport tests pass on the current base.
• Final full package unit suite: 2,147 passed, 7 skipped. Aggregate coverage is 97%; every changed executable line is covered.
• Package-wide Ruff import/format checks (pinned Ruff 0.14.14), flake8, and package metadata/RST checks pass.

The local run covered Python 3.12; the full supported Python/OS matrix and cloud system tests were not run. Existing deprecation and async-mock warnings remain in the package suite.

This change was developed with AI assistance. Reproduction and validation were performed locally without cloud credentials or cloud services.

Thank you for opening a Pull Request! Before submitting your PR, there are a few things you can do to make sure it goes smoothly:

• Make sure to open an issue as a bug/issue before writing your code! That way we can discuss the change, evaluate designs, and agree on the general idea
• Ensure the tests and linter pass
• Code coverage does not decrease (if any source code was changed)
• Appropriate docs were updated (if necessary)

The existing issue predates this patch. Helper docstrings now describe both response interfaces. Compared with unmodified base 3799568 in the same environment, statement coverage increased from 97.2398% to 97.2410%, and branch coverage increased from 95.8998% to 95.9613%.

Fixes #17139 🦕