Your points are spot on, however the analysis engine aggregates or averages track insights across all users who went up that track. If someone is fanging it up a track on the limiter, when usually most will crawl, he (or she) will be an outlier in the data and wont skew the average so much. Hope that makes sense. That being said, the engine is still being refined, and will benefit from having access to a big dataset - which we now have.
Much of the benefit will come from just knowing who went before you - if you know a GQ with 2inch lift and 35s went up no problem you can make a judgement on whether you will struggle based on how you rig compares, regardless of what difficulty the track has been assigned.
On your example I would also argue in low range, hitting the limiter vs crawling doesnt have a massive impact on your actual speed, more your ability to get over obstacles. When looking from a birds eye view your speed wouldnt look TOO different. Happy to hear rebuttals.