Every single feature that is on the unit cost something to implement. In the big scheme of things, this is a minor feature to implement and as the software platform is shared across dozens of models, the end cost would have been pennies per unit if even that much.
I don't agree with your analysis about user error - that's just nonsense and you could argue that point about every single feature on the projector. It certainly won't have been a factor in them deciding whether or not to implement his particular feature. I expect the feature simply was not on the radar otherwise they would have done it. They added masking but not shift, a slightly odd decision? The masking would have been just as much work as the shift, and I'd say that the masking is a far less useful feature. What does anyone ever use the masking settings for?
Personally, I do think it would have been a worthwhile capability as it avoids constant use of the mechanical lens shift for some applications, but it is by no means essential. Why subject some moving parts to wear when a few hundred lines of code could avoid it?
Just think - if they have to repair / replace just a couple of projectors during their lifetime due to wear and tear on the lens shift mechanism, that would have covered the cost of this work!
I accept that 'absolutely nothing' is not true...