Add audio and speech support to Grok adapter

[tombeckenham]

Summary

The @tanstack/ai-grok adapter currently exposes text, image, and summarize adapters, but lacks audio and speech support. xAI's Grok API supports audio capabilities that should be exposed via the SDK's tree-shakeable adapter pattern.

Proposed adapters

Following the pattern established in @tanstack/ai-fal (packages/typescript/ai-fal/src/adapters/):

• grokAudio — audio generation
• grokSpeech — text-to-speech
• grokTranscription — speech-to-text (if supported by Grok API)

Acceptance criteria

• New adapter files added to packages/typescript/ai-grok/src/adapters/
• Adapters exported from packages/typescript/ai-grok/src/index.ts via the /adapters subpath
• Model metadata updated in model-meta.ts for any new audio/speech models
• E2E test coverage added per CLAUDE.md requirements (update feature-support.ts + test-matrix.ts)
• Changeset created via pnpm changeset

References

• Existing Grok adapter: packages/typescript/ai-grok/
• Reference implementation (audio/speech): packages/typescript/ai-fal/src/adapters/
• xAI API docs: https://docs.x.ai/