I once provided an optimized XSL file which showed how to shorten the rather
complex prefix.xsl file. This file was removed to make clear that it is not yet
used. This decision was correct, as the file was unnecessarily contained in the
resulting JAR file.
On the other hand, deleting it was a bit hastily, because still the potential
of it's content was not evaluated or applied.
So I'd propose to check the usefulness of the contained optimization and then
apply it to prefix.xsl if it proofs benefit.