If we route all requests to dependent services though the API layer, we dont have to expose any of them, and we won't need to worry about a reverse proxy.
Simple routing (no path rewrites or reverse proxy required)
"better" security.
easier to eventually add support for other storage layers (Nextcloud/onecloud)
storage access by processor should be done via the Storage API (not directly to the storage container), since we may have additional storage layers (owncloud, nextcloud, Webdav, etc)
If we route all requests to dependent services though the API layer, we dont have to expose any of them, and we won't need to worry about a reverse proxy.