feat(ai-gemini): add Gemini Omni 1.1 Flash (gemini-omni-1.1-flash)

[tombeckenham]

Summary

Add Google Gemini Omni 1.1 Flash (gemini-omni-1.1-flash) to @tanstack/ai-gemini video generation.

Google GA'd this model on 2026-08-27. It is the production Interactions-API video model. The preview id gemini-omni-flash-preview shuts down 2026-09-30.

Daily model-metadata sync does not cover native Gemini video ids (GEMINI_VIDEO_MODELS is hand-curated; veo- is skipped as non-chat). This has to be a first-party adapter change.

Docs: https://ai.google.dev/gemini-api/docs/omni
Model card: https://ai.google.dev/gemini-api/docs/models/gemini-omni-flash

Scope

• Add gemini-omni-1.1-flash to GEMINI_VIDEO_MODELS and GEMINI_INTERACTIONS_VIDEO_MODELS.
• Keep gemini-omni-flash-preview as a deprecated alias so existing callers compile until shutdown.
• Route the GA id through the existing Interactions video path (same as preview). Do not add a second adapter.
• Map response_format.resolution (360p | 720p | 1080p | 4k, default 720p).
• Duration stays 3–10s per call. Clip length / first-last-frame interpolation already work via prompt parts + previous_interaction_id; do not invent a new API for those.
• Update docs (docs/media/video-generation.md), examples/ts-react-media, E2E interactions-video, and the media-generation skill to the GA id.
• Patch changeset for @tanstack/ai-gemini.

Out of scope

• OpenRouter / Vercel Gateway catalog sync (chore: sync model metadata #1264). Those catalogs do not drive Gemini video model-meta.
• Vertex (no Omni ids there today).

Acceptance

geminiVideo('gemini-omni-1.1-flash') type-checks and creates a background Interactions job. Preview id still type-checks.