I own and work in a recording/mixing/mastering studio and we deliver quite a bunch of different masters, depending on how the tracks will be published.
CD files will generally get a neutral master, but it will be pretty loud.
Vinyl will get a reduced stereo image in the bass region, because that might become an issue (physically, while playing back the record).
Streaming platforms are still a mystery to me. They all have different normalization algorithms, which work differently for each music genre and even song structure (a song with a fade-out will sound slightly differently than one without, for example). We tried around a lot (dozens of different masters for Spotify for example), but due to the normalization they apply, it's generally less of a punishment if you publish a record with higher dynamic range.
To comment on the og topic, I still gotta try that with the new system we now have, but I doubt I will hear a difference.
96db dynamic range and a frequency response to 20khz is plenty in my book.
I have tested myself on 320k mp3 vs hi-res in the past, and couldn't even reliably tell those apart (only 2/3 accuracy in a blind test), but that was with my relatively inexpensive home system back then. I'm interested to see if the outcome changes when I repeat this test on my (soon) new headphone rig (Topping D10->Topping L30->HE400i 2020).