The key to understand the technote is the phrase, "as long as the DQ and DQS driver models are the same with the same testload..." This implies that the time-to-vmeas for DQ and DQS is the same. If you then add time-to-vmeas into the equation where Compensated Flight Time = Simulation Time - time-to-vmeas, you get
Setup Time (From Sim) = (DQS Simulation Time - DQS time-to-vmeas) - (DQ Simulation Time - DQ time-to-vmeas)
Then with the assumption that DQS time-to-vmeas = DQ time-to-vmeas those two quantities are subtracted out of the equation and we are left with
Setup Time (From Sim) = DQS Simulation Time - DQ Simulation Time
So it doesn't matter if we call the measurement Compensated Flight Time or just Simulation Time. The measurement is a relative time between the two signals. Compensated Flight time is important when we simulate and measure time delays that we use in timing margin calculations involving Tco (Clock to output time).
The technote is updated with this additional information to clarify the subject.