@tanstack/ai-openrouter: transcription adapter for /api/v1/audio/transcriptions

[mosaleh-dev]

OpenRouter now serves speech-to-text on POST /api/v1/audio/transcriptions
(e.g. elevenlabs/scribe-v2, openai/gpt-4o-mini-transcribe, openai/whisper-large-v3-turbo),
but @tanstack/ai-openrouter (1.0.1) has no transcription adapter, so generateTranscription()
can't run on an OpenRouter key.

Using createOpenaiTranscription with an OpenRouter baseURL almost works, but its model type
only takes OpenAI ids, and the response's usage.cost is dropped.

Request: a createOpenRouterTranscription(model, apiKey) adapter, like the existing text/image/video ones.
The pinned @openrouter/sdk (0.13.20) already exposes the transcription operation.

Docs: https://openrouter.ai/docs/guides/overview/multimodal/stt