feat: Resilienz (Fallback-Ketten) und Metriken (#5)

- Fallback-Provider (app/providers/fallback.py) fuer STT/LLM/TTS: Provider-Kette
  der Reihe nach; Config *_FALLBACK; build_orchestrator baut Ketten (dedupliziert)
- LLM-Stream-Fallback nur solange kein Token gesendet wurde
- Metriken (app/metrics.py): In-Memory Counter/Timer, keine externe Dependency
- HTTP-Middleware (Requests/Latenz/Status je Pfad); Pipeline-Stufen-Timing stt/llm/tts;
  Fallback-/Fehlerzaehler; GET /api/metrics (JSON + Prometheus)
- Tests: 58 gruen (+6); Doku aktualisiert (README, BEDIENUNGSANLEITUNG, Architektur, .env.example)

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

app/dependencies.py

@@ -19,6 +19,11 @@ from app.providers.llm.openrouter import OpenRouterLLMProvider
 from app.providers.tts.openrouter import OpenRouterTTSProvider
 from app.providers.tts.chatterbox import ChatterboxTTSProvider
 from app.providers.tts.piper import PiperTTSProvider
+from app.providers.fallback import (
+    FallbackSTTProvider,
+    FallbackLLMProvider,
+    FallbackTTSProvider,
+)
 from app.pipeline.input_cleaner import InputCleaner
 from app.pipeline.spoken_response_adapter import SpokenResponseAdapter
 from app.pipeline.tts_normalizer import TTSNormalizer
@@ -198,11 +203,32 @@ def resolve_route(
     return ResolvedRoute(**resolved)
 
 
+_FALLBACK_CLASS = {
+    "stt": FallbackSTTProvider,
+    "llm": FallbackLLMProvider,
+    "tts": FallbackTTSProvider,
+}
+
+
+def _provider_chain(registry, primary: str, fallback_csv: str, module: str, cfg: Settings):
+    """Baut primaeren Provider + optionale Fallback-Kette (dedupliziert, Reihenfolge erhalten)."""
+    names = [primary] + [n.strip() for n in (fallback_csv or "").split(",") if n.strip()]
+    seen, ordered = set(), []
+    for name in names:
+        if name not in seen:
+            seen.add(name)
+            ordered.append(name)
+    entries = [(name, _from_registry(registry, name, module.upper(), cfg)) for name in ordered]
+    if len(entries) == 1:
+        return entries[0][1]
+    return _FALLBACK_CLASS[module](module, entries)
+
+
 def build_orchestrator(route: ResolvedRoute, cfg: Settings = settings) -> Orchestrator:
     return Orchestrator(
-        stt=get_stt_provider(route.stt_provider, cfg),
-        llm=get_llm_provider(route.llm_provider, cfg),
-        tts=get_tts_provider(route.tts_provider, cfg),
+        stt=_provider_chain(STT_REGISTRY, route.stt_provider, cfg.stt_fallback, "stt", cfg),
+        llm=_provider_chain(LLM_REGISTRY, route.llm_provider, cfg.llm_fallback, "llm", cfg),
+        tts=_provider_chain(TTS_REGISTRY, route.tts_provider, cfg.tts_fallback, "tts", cfg),
         input_cleaner=InputCleaner(),
         spoken_adapter=SpokenResponseAdapter(),
         tts_normalizer=TTSNormalizer(),