Using the new feature for base flattening works like a charm, thanks!
However, when I tried to experiment with trenches and wells, I hit several problems.
I use the following record (on Baikonur, near Site 70):
Rect 300 63.0962397 46.0351904 600 500 0 1
When Orbiter puts a hill in these coordinates, and I place a vessel on top of the hill, vessel's coordinates are different:
POS 63.1054616 46.0437039
I suspect my other "flattened" bases are also mismatched - I had to increase radius for the flattening more that I thought would be needed - probably their centres are quite offset.
But second problem is even bigger. Even with dropoff 1, the inclines of the hills (I moved to hills from wells for experimenting, as they are easier to see) are still really shallow - did not get anything sharper than ~45 degrees. For my purposes, I need them - the closer to verticals, the better. Well, correction: I finally did manage to get sharper angles - by elevating the hill to the very impractical several kilometers... leading to the next problem:
Finally, the killer: I could not get hill size parameter smaller than 600 by 500 meters. Even 500 by 500 is just ignored. And the shape of the base of the hill is... well, kinda approximate - and HUGE! For launch pad trenches, the needed precision is down to meters, but, more importantly, size should also be down to meters: silo well is only about 10 meter in diameter (and 40-50 meters deep).
I don't know... I see screenshots of some quite sharp and small dugouts in this thread... I noticed they all seem to be on Moon and Mars. Is it maybe the Earth model being more complex, and getting in the way? I understand that I probably trying to use the feature for what it was not intended for. But for these types of launch facilities on Baikonur, flattening the area is only the "invitation to negotiations": the next "must" step is to prevent flame trenches and silos from "filling with the earth" as surface elevation is loaded...