Unexpected results for Dynamic delay range

In reply to ben@SystemVerilog.us:

Works as per expectation for all 11 scenarios.
One last question :
For the 4 dynamic scenarios ( delay/delay range/repetition/repetition range) is it assumed that the dynamic variables remain constant for a SVA attempt OR can they change during an ongoing attempt ?
To model these dynamic cases wouldn’t the best approach be to capture the dynamic variables during the start of an attempt itself ?