feat: add Voice Pipeline Orchestration for TTS/STT capabilities

- Implement VoicePipeline class with multi-provider support
- Add support for OpenAI, ElevenLabs, Google, and Azure providers
- Include voice activity detection and audio enhancement features
- Create comprehensive documentation and usage examples
- Add full test coverage for all pipeline features
- Enable seamless integration with RealtimeSession via plugin pattern

This feature enables natural voice interactions with flexible provider
options, addressing the 'Future' roadmap item for voice pipeline
orchestration that was marked in the SDK documentation.

Author: Joe-Simo