You are right. I just read the whole PDF (admittedly should've done it before posting, instead of just skimming) and there is no mention of how the pointer displays on the screen.
"The mouse cursor itself became a bitmapped image, and for the first time took the familiar diagonal-pointing arrow shape we know today, as well as morphing into other shapes depending on the task being performed."
From ArsTechnica's article. Though it doesn't say WHY either. http://arstechnica.com/features/2005/05/gui/3/
The alleged "turn it 45 degrees, making it easy to see" rationale is only in surrounding modern text (at these blog posts and the StackExchange answer). It might be right, but if so, it's not supported in the linked 1981 document.