feat(classifier): optional GigaChat-2-Lite hybrid tier classify

LLM classify via LiteLLM gigachat-classifier when heuristic confidence
is low. CLASSIFIER_MODE=heuristic|hybrid|llm. Metrics classifier_source.

router/router.py

@@ -15,7 +15,7 @@ import httpx
 from fastapi import FastAPI, Header, HTTPException, Request, Response
 from fastapi.responses import JSONResponse, StreamingResponse
 
-from metrics import CLASSIFY, DURATION, ESCALATIONS, REQUESTS, metrics_payload
+from metrics import CLASSIFY, CLASSIFY_LLM, DURATION, ESCALATIONS, REQUESTS, metrics_payload
 from orchestrator import Orchestrator, SessionStore, Tier
 from rules_loader import reload_configs
 
@@ -104,6 +104,7 @@ def _router_meta(decision, *, requested: str) -> dict[str, Any]:
         "confidence": round(decision.confidence, 3),
         "requested_model": requested,
         "delegated_internal": decision.delegated_internal,
+        "classifier_source": decision.classifier_source,
     }
 
 
@@ -150,7 +151,7 @@ async def classify_debug(
     body = await request.json()
     messages = body.get("messages") or []
     text = _extract_text(messages)
-    decision = orchestrator.resolve(
+    decision = await orchestrator.resolve(
         messages,
         quality_mode=_quality_mode(x_ai_quality, body),
         session_id=_session_id(body),
@@ -159,6 +160,8 @@ async def classify_debug(
         token_estimate=_estimate_tokens(text),
     )
     CLASSIFY.labels(tier=decision.tier.value).inc()
+    if decision.classifier_source == "gigachat":
+        CLASSIFY_LLM.labels(tier=decision.tier.value, status="ok").inc()
     return {
         "tier": decision.tier.value,
         "lane": decision.lane,
@@ -243,7 +246,7 @@ async def chat_completions(
 
     requested_model = body.get("model", DEFAULT_MODEL)
     if requested_model in ("smart-router", "auto", ""):
-        decision = orchestrator.resolve(
+        decision = await orchestrator.resolve(
             messages,
             quality_mode=_quality_mode(x_ai_quality, body),
             session_id=session_id,
@@ -251,6 +254,8 @@ async def chat_completions(
             has_image=has_image,
             token_estimate=_estimate_tokens(text),
         )
+        if decision.classifier_source == "gigachat":
+            CLASSIFY_LLM.labels(tier=decision.tier.value, status="ok").inc()
         target_model = decision.model
         meta = _router_meta(decision, requested=requested_model)
     else: