That way scripts can have a consitent way to get unique ID key/string, and can use it for access in bpy.data collections as well. Note that this kind of douples the old py API way of doning this (passing tuple of (ID-name, lib-path) instead of just ID-name).