The Unicode standard is huge and there are many .pl files in perl's core dealing with every aspect of Unicode. It's just an optimization that most things are loaded on demand (e.g. if you're using regexps with special Unicode properties (\P, \p) or case insensitive matches or something like lc/uc).

Of course, if you know your data and you're in a mod_perl environment, then preload as much as you can.