We proactively construct a new snapshot immediately after destruction, but this is problematic for two reasons:

• No new storage snapshot actually exists yet
• For simple operations that only ever use a single snapshot, this is wasted work