Yesterday I trained K+B+N vs K.
I downloaded a few chess engines to emulate different resistance.
It took me about 3 hours to master the ending.
2 - 3 hours is a representative time to master any complex endgame position.
There are about 150 basic pawn endings, so it will take 150 x 3 = 450 hours to master them.
My initial estimation of 6 months comes rather close when studying 3 hours a day (150 days ~ 5 months)
Everybody agrees that rook endings are more difficult than pawn endings.
Probably that means that there are more than 150 basic rook positions and not that the average time to master a rook position exceeds 3 hours.
So far I counted 190 basic rook positions, but I haven't a good insight in the whole area yet.
This makes it very clear why endgame study isn't popular at clublevel.
It just takes a serious amount of time.
Lately I compared two kinds of positions (40 in total). Both with 2 rooks and a lot of pawns.
20 positions were rook endings, 20 positions were mate in #x)
At first sight the positions looked equally complicated.
It was striking that the rook endings costed me an enormous amount of effort while I solved most mates within seconds. Clearly a matter of pattern recognition.