“Inconvenient,” perhaps, but not “useless.”
In all seriousness, what is so utterly wrong with just using a subdirectory? Apple’s OS/X system is rather elegant, I think, in taking that approach with how they package all applications: they’re actually folders. Just put your executable together more-or-less as a stub, and put the necessary Perl source modules into a directory in a chosen location.
(When I have done this, I used a nested directory structure: the outer directory contained README.TXT and an inner folder named guts. The entire thing was marked read-only. And the README file contained a message which basically said: “Keep your mitts off that folder and everything that it contains. These aren’t the ’droids you’re looking for.” I got a few whimsical comments about that, so I know that some people read it.) To me, this is every bit as convenient a form of “packaging” as a DLL-file would be, and considerably more maintainable. It might well be more efficient, too, because Windows will routinely use VM-mapped file I/O when working with files.