feat: add max users and realistic streaming tests for Orpheus TTS

feat: add TTFT latency benchmark for Orpheus TTS

feat: add Orpheus TTS continuous batching benchmark

refactor: move wav2lip installer to separate video-avatar repository

feat: add Wav2Lip-ONNX-HQ installer as lightweight alternative to MuseTalk

feat: Add complete microservices architecture for speech-to-speech avatar system

feat: add MuseTalk fast_engine integration and model download script

feat: add warmup to start.sh (5 inferences per service)

Initial commit: dumont-talker speech-to-speech avatar