It's true that the clock is differential, and it is simulated as such. The timing constraints are checked to the differential signal, not the separate legs of the differential pair. Therefore, the diagrams in the timing model wizard show a single waveform that represents the differential clock waveform.
Thank you! Are the signals sampled at both rising or falling edges?
I assume that you mean non-clock nets when you say signals. In that case, yes, signals are measured at both rising and falling edges. Signal transitions are compared to the rising edge of the differential clock waveform to calculate timing requirements.