Hello. I wrote the Reddit thread which AppleInsider eventually picked up on. I'll try explaining what I know so far.
1) It's a high-priority bug, which Apple has been aware of for quite a while. Affects all Macs running the HS + spectre/meltdown patch, but rarely appears in non-T2 macs, which seem to be all affected*
*all apart from iMac Pro, as far as I tested.
2) Personally confirmed the bug on 6 2018 machines so far, not counting the ones friends and colleagues also confirmed, all running 10.14.3, with several different soundcards from different manufacturers (USB2/USB3)
3) The bug is deep inside the kernel, unfixable by interface manufacturers, and is independent of the application that's running and load on the CPU.
the Time and Location daemons are not the cause, they merely trigger the faulty line of code in the kernel and cause pauseAudioEngine().
What's below here is pure conjecture, I am definitely not qualified to speak about the internal workings of MacOS, but I have been reading quite a lot on the topic and I may have found some correlation.
The T2, amongst other things, manages ins/outs, including the USB ports. Which means it also handles USB clocking. I believe there is a hardware issue which introduces significant enough amounts of jitter to cause dropouts when using audio interfaces which transmit audio back to the computer.
There is a lot of confusion regarding USB audio, especially asynchronous/isochronous data transfer, but as far as I understood it asynchronous means the external interface handles the clocking, while isochronous refers to how the data is divided up into chunks while being sent through USB.
Now if the T2 had so much jitter that it couldn't sync up with the PLL in the audio interface glitches such as dropouts would occur.
Wild theory: This only happens on audio interfaces with inputs. Anecdotal evidence has led me to believe the issue doesn't exist on USB DACs which only have audio outs. I have yet to test this on my own, but I plan to shortly.