feat: Jarvis v2 - CPU optimized always-listening for iMac Ubuntu

- PipeWire/pw-record backend (no PortAudio needed) - tested mic live
- VAD: Energy (immediate) + Silero (torch CPU) switchable
- Wakeword: openWakeWord TFLite hey_jarvis 0.4
- STT: faster-whisper base int8 CPU - model loads OK on i3
- SpeakerID: ECAPA embeddings + fallback stats, enrollment store
- Vision: OpenCV Haar + face_recognition + webcam cam0 640x480 verified
- LLM: ollama/qwen2.5:3b, gemini_cli, hermes_api, echo strategies
- TTS: Piper + espeak + echo fallbacks
- Core pipeline: IDLE -> WAKE -> RECORDING -> STT -> SpeakerID/FaceID fusion -> LLM -> TTS -> followup window
- Config via jarvis.yaml, soul.md persona
- CLI tools: run, devices, enroll-speaker, enroll-face, test-vad, test-mic
- Systemd user service ready
- Tested on iMac 2010 i3 550 15GB Ubuntu 26.04 CPU-only

scripts/test_wakeword.py

@@ -0,0 +1,29 @@
+"""Test wakeword detection without full pipeline"""
+from jarvis.audio.capture import AudioCapture
+from jarvis.audio.wakeword import WakeWordDetector
+from jarvis.audio.vad import create_vad
+from jarvis.config import load_config
+import time
+
+cfg = load_config(None)
+vad = create_vad(cfg)
+ww = WakeWordDetector(models=cfg.get('wakeword',{}).get('models',['hey_jarvis']), threshold=0.45)
+
+cap = AudioCapture()
+cap.start()
+print("Say 'Hey Jarvis'...")
+try:
+    while True:
+        chunk = cap.read(timeout=0.1)
+        if chunk is None:
+            continue
+        # Optional VAD gate
+        if not vad.is_speech(chunk):
+            continue
+        det = ww.detect(chunk)
+        if det:
+            print(f"Wake detected: {det}")
+except KeyboardInterrupt:
+    pass
+finally:
+    cap.stop()