Currently when the list command is used with a path to only fetch a part of the meta-data tree, the implementation is not optimal as is reads in the whole tree and extract the path afterwards.
Instead, only the part specified by the path should be fetched and returned, skipping the post processing step.
Additionally a new options should be added for specifying the depth for the tree to fetch. Actually, the tree is always returned until it leafs, even when used with a path (which only influences the root-node). Allowing to skip lower layers is important for clients, which e.g. want only to fetch the available domain names or the available MBean names (like for UI tree browser).