Allow local data to be stored on S3 instead of local disk

[mellowagain]

Shiro stores tons of data locally on disk such as beatmaps (.osu) and replays (.osr). Thanks to low-cost storage services like Wasabi, Vultr, MinIO, Scaleway, Yandex Cloud, DigitalOcean, Dreamhost or Exoscale there is less and less incentive to locally store files (with a backup solution and everything) instead of just letting the cloud manage it.

Shiro should integrate data collection using a generic S3 API requester so S3-compatible hosts such as AWS, Scaleway and Wasabi can all be used interchangebly without a problem.