It’s to avoid malware domains and similar applications of various malfeasance. In short, because they can be used to obfuscate the actually intended url copied, such as fасеbook <— this has Cyrillic characters in it you can’t otherwise notice.
When visiting the page (at least on Firefox), the location bar displays the intended characters, but when copying it you get the percent-encoded form. If the reason for percent-encoding was just to make scams more obvious, shouldn't the form shown in the browser interface also be URL-encoded?
By the way, it seems like that is a much stronger argument when applied to domian names than when applied to the part of the URL after the slash.