I would really like to know more about this, and I can't seem to find anything about it. Are these projects implementing their own drivers for all the various audio hardware they can use instead of using ALSA? That certainly wasn't the case some years back when I used OpenElec.
"ALSA" has two meanings - there is only one set of Linux kernel audio drivers, and they're all ALSA no matter what distro you use. I think you're talking about this meaning.
The other meaning of "ALSA" is libasound / asound.conf / dmix, the old userspace audio stack, which is deprecated since Pulseaudio except for a few holdout distros. ALSA had difficulty mixing multiple playback streams, let alone bitstreaming DTS. I think the GP is talking about this meaning.
I sort of disagree. ALSA didn't really have difficulty mixing multiple playback streams unless you go back really far. dmix works quite well. It has a lot of problems, particularly dealing with routing, hot plugging, etc. but the dmix stuff always worked more or less fine for me, as long as I had it configured. otoh, pulse took quite a while to mature in my opinion, and had plenty of issues with this basic functionality over time, though I will at least be fair and say that it got blamed for a lot of ALSA driver issues and other nonsense too. (I think up until recently it was defaulting to doing resampling with a Speex resampler that suffered from fairly noticable aliasing in many cases, as a result of the projects paranoia to be seen as a resource hog...)
That and bitbanging audio through GPIO or otherwise bypassing the ALSA stack doesn't seem all that unreasonable. I can't find any specific information about it with LibreElec/etc. but it would not surprise me.
Oh well. I think Pipewire has a lot of potential today. It brings much improvement to routing and session management for audio and the hybrid scheduler design and considerations for pro audio use cases makes it very promising. I hope the need to bypass the Linux audio stack can disappear.
On the contrary, that's super easy with ALSA. Literally all you need to do to "bitstream" is "play" the data as if it were audio, and not have any software in between trying to adjust volume or otherwise modify it. You could do it with nothing more complicated than "aplay", making sure to specify the correct card and not dmix.
AFAIK pulseaudio, PipeWire and Jack are all built on top of libasound's snd_ APIs, it's the only sane way to use ALSA - just like you aren't going to make manual syscalls for read or write but going to use a libc
They run their own audio stack.