Support for using S3 as export facility

[novoj]

Currently we support exporting files (backups and all other files) to a local filesystems. When running in a cluster environment it's generally recommended to store data to a network operating S3 implementation. The benefit of this approach is to be able to access the exported data from all across the cluster pods.

This mechanism could be also used by replication facility to create snapshot of the actual data in regular intervals which could be downloaded by boostrapping pods to quickly initialize their data set, on which they'd replay missing part of the WAL.

Note - S3 client: https://github.com/minio/minio-java