A possibility that could be explored is to only ship development packages for the wide libraries. Doing so would give automatic UTF-8 support at the cost of increased memory usage, see for instance https://bugzilla.redhat.com/show_bug.cgi?id=1270534#c26.