feat(01-02): LLM Backend Pool — LiteLLM Router with Ollama + Anthropic + OpenAI fallback

- Create llm_pool/router.py: LiteLLM Router with fast (Ollama) and quality (Anthropic/OpenAI) model groups
- Configure fallback chain: quality providers fail -> fast group
- Pin LiteLLM to ==1.82.5 (avoid September 2025 OOM regression in later releases)
- Create llm_pool/main.py: FastAPI service on port 8004 with /complete and /health endpoints
- Add providers/__init__.py: reserved for future per-provider customization
- Update docker-compose.yml: add llm-pool and celery-worker service stubs

packages/llm-pool/pyproject.toml

@@ -9,7 +9,10 @@ description = "LLM Backend Pool — LiteLLM router for Ollama, vLLM, OpenAI, Ant
 requires-python = ">=3.12"
 dependencies = [
     "konstruct-shared",
-    "litellm>=1.54.0",
+    # Pinned: do NOT upgrade past 1.82.5 — a September 2025 OOM regression exists
+    # in later releases. Verify fix before bumping.
+    "litellm==1.82.5",
+    "fastapi[standard]>=0.115.0",
     "httpx>=0.28.0",
 ]