The "inaudible" low frequencies are absolutely removed during mixing, or by the mastering engineer. There's probably no content below 50hz on anything you listen to. Few speakers can reproduce such low sounds (and fewer still do it with any sort of fidelity, and fewer of those are in rooms that won't freak out at the low frequencies). But the amp and speaker use enormous power trying to reproduce them! It seriously injures the sounds we can actually hear to reproduce the sounds we can't hear.
Amps and speakers are the real limiting factors, not the recording medium. They distort very easily (and never read the distortion figures of amps - they're not anything like real world conditions). Mixing is done to make the speakers work well, not to be "accurate".
edit: When tracking and mixing records, I routinely put a high-pass filter on every single track, specifically to prevent subsonic rumble. I often low-pass tracks as well, to limit noise, hiss, and distortion artifacts. This is so routine that even many microphones have a built-in high-pass switch. In mixing, I might high-pass guitars as high as 300hz, throwing away whole octaves of low-frequency information. This is to keep it from conflicting with kick drums, bass guitar, and other instruments that need that space. Solo the guitar in the mix, and it sounds weird. In context, you can't hear the difference in the guitar, and the mix itself sounds better.
Is this a more modern practice, in the same vein as the whole compression thing or would it be something they've been doing all along? I always liked the idea of the "Naked Sound" though perhaps this hasn't ever been a thing ...
No, it's always been the practice, since the 1950s at least. Microphones and media that could reproduce far more accurately than amps/speakers have been around since at least the 1940s. Modern mixing developed in the mid/late 1960s and into the 1970s.
Mixing any sort of multitracked pop music is extremely artificial. Consider any record where an acoustic guitar and an electric guitar are roughly the same volume. In practice, you can't even hear an acoustic guitar next to a raging guitar amp! In the stereo field, things are panned hard left, hard right, or center (there's even a word for it - LCR. Off-center panning sounds weird). Radical eq alterations, compression, multi-miking techniques... heck, move a microphone an inch relative to an acoustic guitar, and it can sound completely different. I have started using two mics on acoustic guitars - one about a foot away, and the other six or more feet away, and blending them. The blend sounds better than either mic by itself. And then there are synths, drum machines, and other direct electronic instruments...
Yeah, "naked sound" in recording doesn't exist. I've been playing acoustic guitar for over 30 years, and play drums and sing and do a lot of live folk music. Recorded/amplified music sounds VERY different. And it should.
Amps and speakers are the real limiting factors, not the recording medium. They distort very easily (and never read the distortion figures of amps - they're not anything like real world conditions). Mixing is done to make the speakers work well, not to be "accurate".
edit: When tracking and mixing records, I routinely put a high-pass filter on every single track, specifically to prevent subsonic rumble. I often low-pass tracks as well, to limit noise, hiss, and distortion artifacts. This is so routine that even many microphones have a built-in high-pass switch. In mixing, I might high-pass guitars as high as 300hz, throwing away whole octaves of low-frequency information. This is to keep it from conflicting with kick drums, bass guitar, and other instruments that need that space. Solo the guitar in the mix, and it sounds weird. In context, you can't hear the difference in the guitar, and the mix itself sounds better.