sachistu: Am curious <sneaky pete> where you found the information about the draw being prearranged?
Actually, this was not the last round game. It was from round 14. There were 15 rounds (with one bye). Going into round 14, Pfleger had only 6.5 points (by my count), so he would have had to at least drawn with both Petrosian and Ljubojevic (his 15th round opponent). Since he had already lost 3 games, that seems hardly guaranteed.
It's unfortunate Pfleger was not more detailed about the situation (Schach Echo November 1974 cover page) wherein he offered the draw. Given his score at the time, it would be surprising he would refrain from playing 17.g4!
Interestingly, both Chess Player 7 (p.240) and Shekhtman's book (Vol 2) on Petrosian give the 12...Nh5 version and without comments. However, it's hard to believe Petrosian would have ignored the danger to his Knight (prearranged draw or not).
Frankly, the score as given (e.g. 12...Nh5) looks suspect. Why would Petrosian give up on his ...b5 break, and even if so, why play the QN to a6 and then back to c7 a second time?! The maneuvre ...Nh6-f7 (rather than ...Na6/c7) seems more in keeping with Petrosian's style. Of course, Petrosian would have wanted to avoid the positionally weakening ...f5, but it would have been necessary in the version given here.
The line/version <stanleys> attributes to Yanvarev is given in the 2016 version of Chessbase. This line seems much more plausible.
The move 12...Ng4 has been played several times as well as 12...Nfe8 and even 12...Nd7. The move 12...Nh5 does not appear (other than here), and rightfully so.