I did find my old dump as you run code, but it has issues. As expected, changes to the Nintendont code base have broken it, so I'd have to refix it to work with current. Even if I fixed it however, I only ever got it to work partially. My Action Replay works enough to dump the whole thing, but the AGP disc does not. I now remember that I used it to partially dump the disc, then used Dolphin to finish getting the rest of the offset/lengths. I would need to add some more patches to get it working the rest of the way. I also never finished getting the EXI (memory card working). If this is still wanted, I guess I could keep working on it. It's in pretty rough shape now (coding/debugging without a USBgecko was a bit of a mess. I have a Shiruken now, but I've never used it, and would face a learning curve if I wanted to).
My suggestion is still using Dolphin. I modified my dumping code slightly.
It now has a default behavior, and generates a .skp file with the crc100000 and skip offsets. If a crc100000=default setting exists, it will use that for skip information. If not, it will attempt to go through the whole disk, and will skip ahead to the next 0xX0000000 offset when finds an error (I left commented out lines for both 0xXX000000 and 0xXXX00000).
Doing this with both my discs found that from 0x00100000-01F00000, the disc is readable and is filled with 0xA8, which is obviously different than the 0x00 fill value I used everywhere, but I don't believe either disc ever actually reads that range.
Other than that range, using the 0xX0000000 offset produced the same data for my Action Replay, but not the AGP. Switching to 0xXX000000 didn't add anything, but took the dumping time from ~10 minutes to ~20 minutes (88 available retry offsets). Switching to 0xXXX000000 would presumably work for my AGP disc, as all the offsets start with those offsets, but would take several hours and would give the disc motor quite a workout (~1300 available retry offsets). I don't know if there's a faster/less motor intensive way to reset after an error. The code should work though if someone wanted to try it.
My recommendation is to use one of the shorter (0xX0000000 or 0xXX000000) ones to get a starting image, and then use Dolphin to get any missing offsets (note that Dolphin requires the IPL file to work with Datel discs).
As an aside, it would be nice to know if all the Datel discs use GNHE as their identifier. The NHL Hitz 2002 info in the datel.dat file is there to support auto Datel detection, but I wasn't sure if they were all GNHE (or possibly GNHP?).