fix(p0): 實測修正 — timeout 依 benchmark 調整，_local_fallback_chain 移除雲端 Nemotron

- config.py: NEMOTRON_DIAGNOSE_TIMEOUT_SECONDS=60s (NIM 實測 11-45s + 15s buffer)
- config.py: OLLAMA_DIAGNOSE_TIMEOUT_SECONDS=200s (Ollama 實測 ~173s + 27s buffer)
- ollama.py: 新增 per-task timeout (diagnose/force_local 用 200s)
- ai_router.py: _local_fallback_chain 移除 Nemotron (NIM=雲端，不可進 local chain)
- ai_router.py: v4.2 — Option C 分情境路由正式確立

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

apps/api/src/services/ai_providers/ollama.py

@@ -71,6 +71,15 @@ class OllamaProvider:
             model_name = registry.get_model("ollama", "rca")
             options = registry.get_provider_options("ollama")
 
+            # P0 2026-04-04 Claude Code: per-task timeout（Option C 分情境）
+            # FORCE_LOCAL/diagnose → OLLAMA_DIAGNOSE_TIMEOUT_SECONDS (200s，實測 ~173s)
+            # 其他 → OPENCLAW_TIMEOUT（既有設定）
+            task_type = (context or {}).get("task_type", "")
+            if task_type in ("diagnose", "force_local"):
+                read_timeout = float(getattr(settings, "OLLAMA_DIAGNOSE_TIMEOUT_SECONDS", 200))
+            else:
+                read_timeout = float(settings.OPENCLAW_TIMEOUT)
+
             response = await client.post(
                 f"{settings.OLLAMA_URL}/api/generate",
                 json={
@@ -84,7 +93,7 @@ class OllamaProvider:
                         "top_p": options.get("top_p", 0.9),
                     },
                 },
-                timeout=httpx.Timeout(float(settings.OPENCLAW_TIMEOUT), connect=10.0),
+                timeout=httpx.Timeout(read_timeout, connect=10.0),
             )
             response.raise_for_status()
             data = response.json()