To me it’s nothing different than brushing return with a fake a higher return number, so why it’s legitimate??
Maybe the point is that excluding outliers sometimes gives you a more meaningful measure for analysis. But it can also be exploited (I guess by outright misrepresentation in the reporting? Not clear about that), just like you can always play with the numbers until they look good.
You are right, for me I omitted the word “outliers” in the first line so removing the extreme numbers sound reasonable and even necessary which gives some room to play.