AbstractVoice v0.8.5

Added

• Remote OpenAI/OpenAI-compatible voice engines for VoiceManager: tts_engine="openai"|"openai-compatible" and stt_engine="openai"|"openai-compatible" call /audio/speech and /audio/transcriptions directly without importing AbstractCore.
• Remote-compatible cloning support through cloning_engine="openai"|"openai-compatible" with a configurable clone endpoint and stored remote voice ids for later synthesis.
• Remote/provider voice profiles: OpenAI built-in voices and compatible endpoint voice lists are exposed through VoiceManager.get_profiles(), selected with set_profile(...), and sent as the remote speech voice.
• REPL and web UI startup flags for remote engines, remote models, base URL, API key, and timeout.
• Local web compatible extension routes: GET /v1/audio/voices for profile/voice discovery and POST /v1/voice/clone for remote clone creation smoke tests.

Changed

• The local web /v1/audio/speech alias can treat voice as either a cloned voice id/name or an active-engine profile id.
• tts_engine="openai" and stt_engine="openai" now default to OpenAI's hosted https://api.openai.com/v1 and read OPENAI_API_KEY.