Well, yes, usually the extra stack work wouldn't amount to much, unless/until the alias in question ends up getting called two million times on a given data set...

Wouldn't that be a fun and enlightening exercise in benchmarking? Someone uses the same module with the same test code on the same data set, but just switches between the "alias" name and the "real" name of the function, and gets to see the impact of the extra stack layer. (But I can imagine that some folks would feel like they had better things to do with their time.)