Every paper I’ve looked at treats the cessation of admixture as the beginning of endogamy. I’m a novice in this aspect of genetics, my opinion on whether this conceptually makes sense or not isn’t relevant. I’m deferring to what the experts seem to be saying.
The median North Indian endogamy time from the two studies discussed in the article, is around 2200 years ago. The average for Punjabis in the new study is around 1300 years ago. That’s a significant difference, and to argue it is not would imply these studies are at baseline worthless, if nearly 1000 years can be considered within error parameters.
Once again, the authors seem to disagree with this idea. The Basu study treats a 580 year window between North Indians and the Marathas as significant, discarding the Gupta model in light of the later endogamy date for this southern population, and instead chalking it up to the Chalukya and Rashtrakuta polities of the Deccan.