It is popular indeed. Consider:
3B4/1r2p3/r2p1p2/bkp1P1p1/1p1P1PPp/p1P1K2P/PPB5/8 w - - acd 49; bm Ba4+; ce 0; c0 "tt06 {m}; 1912; Rudolph W; Draw;"; c1 "Hard-Talkchess-2020.054"; c2 "W.E. Rudolph, La Stratégie 1912 (William E. Rudolph of Brooklyn, NY page 184 of the March 1911 Chess Amateur) White to play and draw"; c3 "Ba4+"; c4 "Rudolph, W.E_?_Studie 1912 HTC108-30 S, 164"; c5 "Rudolph=W; 1/2"; c6 "Arves.06200"; c7 "The Chess Amateur#0181;1911;W:Rudolph=W;(=0650.88e3b5);1/2-1/2;1911"; c8 "pv Ba4+ Kxa4 b3+ Kb5 c4+ Kc6 d5+ Kd7 e6+ Kxd8 f5 1/2-1/2"; pm Ba4+; pv Ba4+ Kxa4 b3+ Kb5 c4+ Kc6 d5+ Kd7 e6+ Kc8 f5 Kxd8 Kd2 Kc7 Kd1 Rb8 Kd2;
As you can see, there are references to may test sets and publications noted within the data record. Besides all that it is problem id "MES0941" from the MES test set, and it is found in YACPDB as item number 289349. It is further found as P1269040 on the Schwalbe PDB database server.
The thing that makes this position interesting is that in 2023, with enough horsepower to gob-smack deep blue, many top engines still do not understand the position. Some get the wrong answer altogether. Some will get the right answer but show a huge negative score.
Does that seem interesting to you? Because it is interesting to me.
The original poster was not trying to solve the position to find the right answer. Everyone knows the right answer. The point was to see how Stockfish did against the position, because formerly it did terrible. It now does some better, but does not understand that the position is drawn.
There are engines with special code that will find the correct answer with a score of zero. But that same detection code makes them run slower for other problems.
So the most interesting question (it seems to me) is "Can engines find the right answer with the right score and do it efficiently?"
That is the open question that makes the position worthy of further study. IMO-YMMV