Feature: Introduce JUCE-Based DSP Engine for Client-Side Voice Processing

### 🚀 The feature

Add a shared, DSP-only audio processing engine using JUCE (C++) to enable real-time, client-side voice transformation.
The engine will support predefined voice profiles (e.g., pitch, filter, gain) and will be designed to run safely in real-time audio threads. The DSP logic will be platform-agnostic and reused across Android and iOS.

No UI, networking, or platform-specific logic will be included in this component.

### Motivation, pitch

Currently, audio published to LiveKit uses raw microphone input. Introducing a JUCE-based DSP engine enables optional, privacy-preserving voice transformation entirely on-device, without any server-side processing.

JUCE is a well-established real-time audio framework and allows us to implement DSP logic once in C++ while sharing it across platforms. This keeps the architecture clean, performant, and maintainable while unlocking new audio features such as voice styles and masking.