In case you didn't notice, I referenced the directivity index and the heat map.
the directivity index does not matter as much as you think it does, the early reflection directivity index (ERDI, dotted blue below DI) is much more important and it's fairly smooth as well. When you mentioned 'people sitting off-axis will experience different things' well that's where you're wrong because ERDI is smooth, and that's the measure used to reflect the consistency of the off-axis response.
The predicted in-room response is 12% Listening window, 44% Early reflections, and 44% Sound power/ Power response. So a little bit messy DI is almost insignificant when ERDI is fairly smooth.