I don’t know anything about your hardware but a couple of thoughts at you…
The bottom line - Don’t lose sleep over it! (Except, of course, for the 1 or 2-second gaps, etc. 
)
“CD quality”, which is 16-bits, 44.1kHz, stereo, is generally better than human hearing. You won’t hear a difference between a high-resolution original and a copy downsampled to 16/44, in a proper, blind, level-matched ABX Test. People “hear” all kinds of differences in non-blind tests. 
Sadly, “the audiophile community” is mostly nuts! And most audiophools “don’t believe in” blind listening tests.
As you may know, MP3 is lossy compression and data is thrown-away to make a smaller file. But a good-quality MP3 can often sound identical to the uncompressed original (in a blind listening test) or you may have to listen VERY carefully to hear the difference. There ARE, of course, low-bitrate-low-quality MP3s and MP3 is ALWAYS lossy.
If you have a high resolution file (or stream) and your hardware supports that resolution, there’s no reason NOT to use it. Unless, sometimes you may want to convert or compress to make a smaller file.
Note that FLAC is lossless compression. It can make a file almost half the size of the original without throwing-away any data. (Similar to making a Zip file.)
Up-sampling doesn’t make the audio better. It’s like copying a VHS tape to Blu-Ray… You don’t magically get Blu-Ray quality.
If you’re on a computer, the drivers will quietly downsample to match your hardware if necessary. i.e. You can play a high resolution file on any-old cheap soundcard just like you can print a high-resolution photo on a low-resolution printer and the operating system won’t warn you.
Personally, I’ve never heard MQA. It’s supposed to be some sort of “enhancement”. It’s not in the original recording from the studio. I think it’s subtle if you can hear it at all, and there are probably better, more straightforward, ways to “enhance” the sound (EQ, etc.).
There are no 32-bit analog-to-digital converters. Again the software/drivers will make a conversion (up or down) to match the bit-depth of your hardware.
Audacity uses 32-bit floating-point internally (like most audio editors/DAWs) because the digital signal processing is “easier” with floating-point and you can temporarily go over 0dB without clipping.
At 8-bits you can hear quantization noise. It’s like a “fuzz” on top of the audio. Like regular analog noise it’s most-noticeable with quiet sounds but it goes-away with digital silence. The high frequencies are limited to half of the ample rate. If you downsample to 8kHz, the audio only goes to 4kHz . AT 44.1kHz the audio can go above the “traditional” audible range of 20kHz. (Usually, only young people can actually hear to 20kHz.)
NOTHING in the digital domain (or in the electronics) will accidently affect that. Of course you can intentionally boost the bass, etc. The frequency response of speakers (or headphones) makes a BIG difference in sound character/quality.
There are only 2 speakers (with stereo) and the sound comes from the speakers so “soundstage” is an illusion, mostly from the recording, the room, the speakers, and the listener’s perception. (With headphones, most people hear the soundstage coming from inside their head!)
See Audophoolery.