fix(core): stop blocking the event loop in OpenAI parsing and static routing

OpenAIParser now uses AsyncOpenAI — the previous sync SDK call ran directly
on the event loop (submitting to a ThreadPoolExecutor and immediately
blocking on .result() serialized it anyway). TitleParser.raw_parser and its
callers become async accordingly.

The SPA catch-all route listed dist/ on every request; snapshot it once at
startup. The module-level bangumi TTL cache is now lock-guarded — it is
written from asyncio.to_thread workers in notification paths while the event
loop reads it.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014w1Z6Nxy6XTRgkFXqPr9Zh

backend/src/module/parser/title_parser.py

@@ -57,14 +57,14 @@ class TitleParser:
             logger.warning("Please change bangumi info manually.")
 
     @staticmethod
-    def raw_parser(raw: str) -> Bangumi | None:
+    async def raw_parser(raw: str) -> Bangumi | None:
         language = settings.rss_parser.language
         try:
             # use OpenAI ChatGPT to parse raw title and get structured data
             if settings.experimental_openai.enable:
                 kwargs = settings.experimental_openai.dict(exclude={"enable"})
                 gpt = OpenAIParser(**kwargs)
-                episode_dict = gpt.parse(raw, asdict=True)
+                episode_dict = await gpt.parse(raw, asdict=True)
                 episode = Episode(**episode_dict)
             else:
                 episode = raw_parser(raw)