So, my question is: How can I make the older versions of HTML::Parser treat unicode entities?

Why do you insist on using an old version? Is the module etched into the motherboard of the computer? New versions are made because there's something wrong or suboptimal about the old version, and in general you're supposed to upgrade in order to take advantage of the changes.