feat(api): sync test API (serve) + opt-in LLM correction + cloudflared tunnel

- api/: FastAPI app, X-API-Key 인증(미설정 시 임시키), 엔진 load-once 풀
  (+transcribe lock), POST /v1/transcribe(multipart, 동기), /health, /v1/system,
  /v1/models. 업로드 임시파일 finally 삭제(프라이버시).
- postprocess/: llm.correct(scripts/llm_correct.py 승격; opt-in·allowlist·감사로그·재시도)
  + rules.normalize(EmbeddingGemma 등 정규화).
- results/formats.py: txt/srt/vtt. connectivity/tunnel.py: cloudflared quick tunnel(Colab).
- cli serve: uvicorn 단일워커 + --tunnel cloudflare; config llm_* 필드;
  pyproject api/queue extra 분리(+python-multipart, dev httpx).

검증: 22 단위테스트(API TestClient·formats·postprocess) + 실서버 e2e
(/health·auth 401·실제 전사(JFK)·SRT·임시파일 삭제). KO 품질은 turbo/large-v3 필요(tiny는 한국어 degenerate).

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

tests/test_postprocess.py

@@ -0,0 +1,41 @@
+"""postprocess.rules / postprocess.llm (urllib monkeypatch)."""
+from __future__ import annotations
+
+import json
+
+import pytest
+
+from luke_scribe.postprocess import llm, rules
+
+
+def test_rules_normalize():
+    assert rules.normalize("구글 Embedding Gemma 소개") == "구글 EmbeddingGemma 소개"
+    assert rules.normalize("그대로") == "그대로"
+
+
+def test_llm_not_configured():
+    with pytest.raises(llm.LLMNotConfigured):
+        llm.correct("x", base_url=None, api_key=None)
+
+
+class _FakeResp:
+    def __init__(self, payload: dict) -> None:
+        self._p = payload
+
+    def read(self) -> bytes:
+        return json.dumps(self._p).encode()
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, *_a):
+        return False
+
+
+def test_llm_correct_monkeypatched(monkeypatch):
+    def fake_urlopen(_req, timeout=90):  # noqa: ARG001
+        return _FakeResp({"choices": [{"message": {"content": "EmbeddingGemma 복원됨"}}]})
+
+    monkeypatch.setattr(llm.urllib.request, "urlopen", fake_urlopen)
+    out = llm.correct("인베딩 점마", base_url="http://x/v1", api_key="k", model="m")
+    assert out == "EmbeddingGemma 복원됨"