Process Drive folder uploads asynchronously in a worker

[ademboukabes]

Summary

This PR moves Google Drive folder imports out of the request/response cycle and processes them asynchronously in a dedicated worker.

What changed

• added a dedicated upload_group_worker
• added a new NATS subject for upload group import requests
• changed folder-based upload creation to return immediately after creating the upload group
• moved heavy Google Drive folder import work to the worker
• added upload group processing fields:
  • processing_status
  • processed_photo_count
  • failed_photo_count
  • error_message
• updated upload group response schema to expose processing progress
• added the migration for upload group processing state
• added the new worker to run-workers
• fixed storage_cleaner settings for Pydantic v2
• fixed worker bootstrap issues by initializing Redis and MinIO properly

Why

Previously, folder-based Drive imports were processed synchronously inside the HTTP request, which made large folder uploads slow and blocking.

Now the API returns quickly, and the import is handled in the background by a worker.