feat: enable orchestrator + switch TTS to Edge (Katja voice)

- Enable delegation.orchestrator_enabled with max_spawn_depth=2
- Switch TTS from ElevenLabs (paid) to Edge TTS (free) with de-DE-KatjaNeural
- Katja is a warm, friendly German female voice — no API key required

hosts/m3-hermes/services/hermes-agent.nix

@@ -4,8 +4,8 @@
   inputs,
   ...
 }: let
-  # Default ElevenLabs voice: Bella (German-capable female)
-  elevenlabsVoiceId = "hpp4J3VqNfWAUOO0d1Us";
+  # Edge TTS: Katja — warm, friendly German female voice (free, no API key)
+  edgeVoice = "de-DE-KatjaNeural";
 
   # Extra Python packages from the container's writable venv layer.
   # matrix-nio is installed via pip in /home/hermes/.venv but the hermes
@@ -180,10 +180,9 @@ in {
 
       # ── TTS / STT / Voice ──────────────────────────────────────────────
       tts = {
-        provider = "elevenlabs";
-        elevenlabs = {
-          voice_id = elevenlabsVoiceId;
-          model_id = "eleven_multilingual_v2";
+        provider = "edge";
+        edge = {
+          voice = edgeVoice;
         };
       };
 
@@ -209,9 +208,11 @@ in {
         user_char_limit = 1375;
       };
 
-      # ── Delegation ─────────────────────────────────────────────────────
+      # ── Delegation / Orchestrator ────────────────────────────────────────
       delegation = {
         max_iterations = 50;
+        orchestrator_enabled = true;
+        max_spawn_depth = 2;
       };
 
       # ── Matrix ────────────────────────────────────────────────────────