IIUC, the salient points are: - target area was very small (500m x 500m) for CY2. With CY3 they've made it bigger at 4km x 4km, allows for larger margin of error.

- CY2 lander had limited leeway in fixing issues, by design. CY3 has more and has landed itself, no assistance from base.

- CY2 lander had limited time to fix itself, apparently it was just a few seconds short of making it fine

+ Software limited the upper limits of thrust correction/per second to 10% of what was possible.

In other words, it could perhaps correct itself in those many seconds but was allowed only small degrees of change.

I asked Kagi's Universal Summarizer, and it said this:

The video provides an explanation for why India's Chandrayaan 2 lunar landing mission failed in 2019. The landing craft Vikram experienced issues during the camera coasting phase where instruments were being calibrated. Specifically, the engine thrust was higher than expected and thrust control couldn't be adjusted to correct the trajectory. As a result, errors accumulated and Vikram's trajectory deviated significantly from the planned path. For Chandrayaan 3, ISRO has implemented changes like adding instantaneous thrust control, increasing the allowed attitude change rate, and enlarging the landing zone to make the mission more robust to errors. The video offers an insightful technical look into what went wrong and how ISRO is addressing it for the next lunar mission.