Generally at electrical test you'd run the pattern in a loop
and walk the strobe back from the final value toward zero
until the output has "broken out" of its "accept mask" (or
simple limit). This works in simulation too.
You need to assert a tolerance or U/L limit pair for a quantity
you can get at. Then develop that quantity (e.g. the pulse
peak or flat-top). Then a rack of measure statements to
grab each (say) flat-top voltage into a vector, along with a
companion time-point vector.
Now index back to front through that "accept" vector with
a break-logic using the limit-pair, and report value and time
at the index.