The one time this was a problem for me (thanks bad HP/Compaq server NIC driver!) I found that following the procedure in that article wasn't really all that bad, and quickly zeroed-in on the cuplrit. I'd recommend giving it a shot-- it's really not that bad. Once I had pool tagging enabled and let the machine run for a couple of days I had my answer in PoolMon in just a few minutes.
Edit: That's a pretty old version of iaStor.Sys. I can't find any other reports of nonpaged pool leaks with iaStor.sys, but it's possible.
The current version of iaStor.sys is an "8" version. I'd consider pulling down an update from Intel (http://downloadcenter.intel.com/filter_results.aspx?strTypes=all&ProductID=2101&OSFullName=Windows*+XP+Professional&lang=eng&strOSs=44&submit=Go%21). I can't tell you that there isn't some potential risk in doing such a driver upgrade, but it ought to go smoothly. Decide based on your tolerance for risk and your level of backup of the data if you should proceed with an upgrade.
I'm curious about that "xns2" pool tag. You say that's from USBSTOR.SYS? Did you find that with the "Parse a Driver for Memory Tags" function in PoolMon.exe?