Add MiniMax text-to-speech support

[octo-patch]

Reason: Add native MiniMax text-to-speech support through the existing media provider route.

• Send regional /t2a_v2 requests with the current default model, speech options, and hex or URL response handling.
• Route minimax/ audio models to the media provider while preserving provider validation errors.
• Add focused coverage for request construction, regional endpoints, response parsing, validation, and high-level routing.

Checks:

• uvx --from ruff==0.15.22 ruff check .
• uvx --from ruff==0.15.22 ruff format --check agentfield/media_providers.py agentfield/agent_ai.py tests/test_minimax_audio.py tests/test_minimax_video.py
• uv run --frozen --extra dev ./scripts/run_pytest.sh tests/test_minimax_audio.py tests/test_minimax_video.py tests/test_media_providers.py tests/test_agent_ai.py --no-cov

[github-actions]

Performance

SDK	Memory	Δ	Latency	Δ	Tests	Status
Python	9.0 KB	-	0.30 µs	-14%	✓	✓

✓ No regressions detected

[github-actions]

📊 Coverage gate

Thresholds from .coverage-gate.toml: per-surface ≥ 84%, aggregate ≥ 85%, max per-surface regression ≤ 1.0 pp, max aggregate regression ≤ 0.50 pp.

Surface	Current	Baseline	Δ
control-plane	87.10%	87.40%	↓ -0.30 pp	🟡
sdk-go	92.70%	92.00%	↑ +0.70 pp	🟢
sdk-python	93.82%	93.73%	↑ +0.09 pp	🟢
sdk-typescript	91.08%	90.42%	↑ +0.66 pp	🟢
web-ui	84.76%	84.79%	↓ -0.03 pp	🟡
aggregate	85.62%	85.75%	↓ -0.13 pp	🟡

✅ Gate passed

No surface regressed past the allowed threshold and the aggregate stayed above the floor.

[github-actions]

📐 Patch coverage gate

Threshold: 80% on lines this PR touches vs origin/main (from .coverage-gate.toml:thresholds.min_patch).

Surface	Touched lines	Patch coverage	Status
control-plane	0	—	➖ no changes
sdk-go	0	—	➖ no changes
sdk-python	0	—	➖ no changes
sdk-typescript	0	—	➖ no changes
web-ui	0	—	➖ no changes

✅ Patch gate passed

Every surface whose lines were touched by this PR has patch coverage at or above the threshold.

[AbirAbbas]

Took this over along with your other two MiniMax PRs since they'd gone quiet.

This was the cleanest of the three. I ran TTS end-to-end against a mocked transport and confirmed the /t2a_v2 request shape for both the global and China endpoints, both the hex and URL response paths, and every validation branch — all behaved correctly. I also checked the agent_ai.py routing change specifically, since that touches the public high-level API: minimax/ audio models reach the media provider and nothing else changed route.

On the test_minimax_video.py edit — I checked it isn't weakening anything. It only removes the now-obsolete assertion that generate_audio() raises NotImplementedError, which this PR makes untrue. Fine as-is.

Two small things I pushed a commit for:

The ai_generate_audio docstring still listed only LiteLLM and Fal.ai and documented only OpenAI voices, so there was no way for a caller to discover MiniMax support or realize the alloy default doesn't apply to it. Updated that and ai_with_audio with a minimax/speech-2.8-hd example.

The validation test only covered four of the branches this PR adds. I verified the uncovered ones behave correctly by hand, but a regression in any of them would have passed the suite — so I parameterized it to cover invalid output_format, speed-without-voice, missing voice_id, empty model, HTTP and API errors, malformed JSON, incomplete status, absent audio, and invalid hex. No production change was needed for that.

Heads up that #853 conflicts with this one in supported_modalities and the shared video test. This one goes first; I'll reconcile #853 on top.