Support fetching world and other init data from S3

### Enhancement Type

Improve an existing feature

### Describe the enhancement

It would be great if all features that support downloading from a URL or container path, e.g. WORLD, OPS_FILE, WHITELIST_FILE, ICON, etc. supported getting the respective files from a (private) S3 or bucket or other compatible object storage.
An easy workaround is proxying via nginx-s3-gateway or similar, but builtin support seems more straightforward.

Thanks for considering!

Start by tracing the existing startup handling for WORLD, OPS_FILE, WHITELIST_FILE, ICON, and other URL or container-path inputs. Identify the entry point and current download logic, then determine how private S3 credentials and compatible object-storage URLs should be supplied. Done means the supported init-data features can fetch from private S3 or compatible storage without requiring an external proxy.

Written by the indexing model from the issue text.