Restores are broken in managed OLTP, and they have been broken for a long time. They are slow, and they get slower with scale. The databases that need recovery most, the large production ones, are the ones left waiting the longest.