I don't have the numbers/measurements right here.. But the delay time is set on the HF so it matched the impulse response measurement of the LF in the listening position. I would think that this is the way to look at it?
I use 24db/oct BW slopes, and set the crossover at 800hz. I'm only measuring from 80 to 8khz because the measurements are regarding the crossover-point. I measured the LF alone, and there is no suckout in the range 600-900hz... If a room suckout was present, this would show in the measurement of the LF alone. The suckout is only present when both drivers are playing together. I have setup everything from information I could find on various guides on the Internet, so suggestions are always welcome.