That line seems very dangerous for white and it's exactly the type of position every king's Indian player wants.
King's Indian Mar del Plata 9. Nd2 busted?!

yeah but im not sure why the op was calling Nd2 refuted without exploring anything other than a few GM games. improvements can be made and it seems like white breaks through before black with these improvements

if you truly want a refutation to the 9.Nb2 line then 9.a5 may be the best idea for black to halt whites queenside expansion for awhile
Rather
King's Indian Mar del Plata busted by 9 b4.
https://www.chessgames.com/perl/chessgame?gid=1070932

Rather
King's Indian Mar del Plata busted by 9 b4.
The Bayonet does not bust the Kings Indian. Players have found solutions for Black since the scare by Kramnik in his win over Kasparov in the 90s.

perfectly playable for both sides. chessbase calls this +0.1 while others say around +0.3-+0.4. black attempts to stop whites queenside play so he can start kingside play. white is okay with this and his rook patrols the 3rd rank as well as active central knights
@50
That is just winning for white
Kasparov gave up on the King's Indian Defence for a reason after his loss to Kramnik.

In sharp lines were black is trouble at a moment in time, can be repaired later. Kasparov games decades ago now, theory has moved on. After 10Re1 black moved away from Nf4, and lots of games with 10...f5 then Ng5. Kramnik's games with this in 90's now viewed old fashioned, and more modern games with 11a4 delaying Ng5. If black likes a5 ideas then 9...a5 probably the way to go, as in Gawain Jones courses/books. Jones also likes 6...Nb-d7 to mix it up a bit too.
@53
"In sharp lines were black is trouble at a moment in time, can be repaired later." ++ How?
"Kasparov games decades ago now, theory has moved on." ++ To what?
"After 10Re1 black moved away from Nf4" ++ What instead of 10...Nf4? 9..Ne8 is busted too. The whole reason for 9 Nd2 was the belief that white had to prevent 9...Nh5 and 10...Nf4.
"lots of games with 10...f5 then Ng5" ++ There black is in trouble too.
"modern games with 11a4 delaying Ng5"
++ Black is in trouble, so we need better moves for black, not for white
"9...a5 probably the way to go" ++ It transposes.
"6...Nb-d7 to mix it up a bit too" ++ 6...Nbd7 or 7...exd4 are alternatives, but that is admitting defeat

@50
That is just winning for white
Kasparov gave up on the King's Indian Defence for a reason after his loss to Kramnik.
these arent the positions i gave you. also the computer even thinks black is fine

Theoretically black is fine after 10Re1 f5 11Ng5 Nf6 lots of Radjabov games particularly against Van Wely were black won a fair number, which is white moved on to looking at 11a4 ideas.
9...a5 doesn't have to transpose to lines in thread. The ones mentioned in thread aren't losing for black either, but it seems most flexible way of playing a5 ideas. Can avoid wasting time with knight with Nh5-f4 -Nh5.
9....Nn-d7 is less forcing and more flexible to my taste. Black just isn't as mathematically lost as seem to think.
white seems to have enough compensation if you follow the line. i dont think 9.Nd2 is busted and is perfectly playable for a win with both black and white