I agree that the fianchetto variations give better winning chances, but probably more losing chances too. Black has to be patient and adept at maneuvering to play the Philidor. In addition its very easy to go very wrong early in the Philidor and get crushed - you have to know what you're doing especially in the 3...Nd7 lines. It's probably best to try a delayed Philidor with the c6, d6, Nd7, Qc7 stuff. Alekhine, who to say the least was a dynamic player, played the Philidor on occasion.
Why not the Philidor?

Regarding the line with 10 Nf5 you are discussing... After 10...Bxf5 11 exf5 Nc6, instead of 12 Bxd6, there was a game Sanchez v Fougertt, 2006, France, which went 12 g4 b4 13 Nd5 Nxd5 14 Qxd5 Bg5 etc I am puzzled by the final diagram position in this game, which is on the database here at Chess.com. It has it down as a white win after 29 Qxf6. Doesn't BLACK have a mate in one?
The first chapter of Chess Strategy in Action by John Watson explains the downsides of the Phillidor--black surrenders the center. Although the opening isn't refuted by any means, black's winning chances mainly stem from fianchetto variations, otherwise black is faced with a troublesome dark-squared bishop.