fix/move orphan check from connection to worker

[stanley2058]

Move the orphan check and recovery to the worker since a large stream count would make Redis halt.

The current process is as follows:

1. When adding a message to a stream, if it exists and its length is greater than 100, the stream would be put into a checklist.
2. When the workers are performing a persisting operation, the one with the smallest ID (sorted by alphabet) is elected as the leader and performs the recovery operation.
   • We only want one worker to do this to avoid race conditions.
3. All workers would maintain a list of streams they are processing in a Redis set to speed up the lookup. The set would also have a TTL to avoid zombie entry if the worker exited.

[Yukaii]

Also getting error when running tests

  @y/redis/api/worker: adding 0 orphans back to worker +104ms
  ClientClosedError: The client is closed
      at Commander._RedisClient_sendCommand (/y-socketio-redis/node_modules/.pnpm/@[email protected]/node_modules/@redis/client/dist/lib/client/index.js:510:31)
      at Commander.commandsExecutor (/y-socketio-redis/node_modules/.pnpm/@[email protected]/node_modules/@redis/client/dist/lib/client/index.js:190:154)
      at BaseClass.<computed> [as expire] (/y-socketio-redis/node_modules/.pnpm/@[email protected]/node_modules/@redis/client/dist/lib/commander.js:8:29)
      at Api.consumeWorkerQueue (file:///y-socketio-redis/src/api.js:288:22)
      at process.processTicksAndRejections (node:internal/process/task_queues:105:5)
      at async file:///y-socketio-redis/src/api.js:369:37