I wonder what might be the motivation behind that. Sounds like a strange policy if they have confidence in their product. Maybe it's for legal reasons.
It doesn't allow them to control the narrative about their product. Suppose the reviewer doesn't know what they're doing, and has the wrong settings selected. Or uses code samples that are devilishly difficult (e.g., heavy use of function pointers) so that the results show the product in a very negative light.
If I build a car and someone shifts the gears without using the clutch, breaks the car and then whines about it, that might be irritating but totally legal. So I wonder if that sort of business practice would hold in front of court.