Hearing your testfiles was interesting, but I found no evidence to prove the difference.. But like someone said, it's easy to compress electronical stuff for the algorithm. Would be way more interesting with vocal stuff!
I believe that during my tests I could every time tell the difference from anything encoded when listening to "white flag" from dido. That Glockenspiel-part sounded different everytime, I could even hear it even on loudspeakers in my living room.
I compared it only to 320 kbps mp3 LAME I encoded years ago from the same CD.
But I believe even AAC wouldn't do magic when it comes to complex scenes, but anyways, what is the cost of storage nowadays anyways? Go lossless all the way! No need for compression and discussions left.... But wait, THEN there are those streaming services right?
In 2014 Spotify sounded muffled all the way through. You didn't even need ABX mechanisms to hear that. Highs were clearly missing. I've then switched to Deezer, which was not lacking of that. AND switched back this year, problem has been fixed, or deafening-age has been reached...
I've seen people that ask, how to achieve that sound, like that they've heard in that sound-studio in tweaking the compression-algorithm. You won't, forget it. You need a set of good speakers, a good amp that goes with it, and then the sound follows!
I'll bet that the day comes, that one streaming service starts to do lossless, and if only one does, everybody will follow for sure....