`AudioRequest` missing `chunking_strategy` field

[alnah]

Is your feature request related to a problem? Please describe.

I'm using gpt-4o-transcribe-diarize for speaker diarization. It works fine for short clips, but anything bigger fails with:

chunking_strategy is required for diarization models: bad request

Looking at the transcription API docs, chunking_strategy is required for diarization on longer audio. But AudioRequest doesn't have this field, so there's no way to set it through the library.

Describe the solution you'd like

Add ChunkingStrategy field to AudioRequest struct.

Also, AudioResponseFormatDiarizedJSON for the diarized_json format would be useful.

Describe alternatives you've considered

Making direct HTTP requests to work around it, which defeats the purpose of using the library.