Cloud storage with file versions and a trash bin

Build an in-memory file store in levels, each adding operations to the last.

1. Add a file with a size (adding an existing name stores a new version), read a file's size (its latest version), move a file with all its versions, and list the n largest files under a path prefix, by size and then name.
2. Read a given version's size, and delete a version permanently; later version numbers shift down, and deleting the only version removes the file.
3. Delete every file under a prefix into a trash bin, and restore every trashed file under a prefix. A name already in the trash, or already active on restore, is replaced by the incoming file.