docs: for now we won't document atomic triples or total triples; de-experimentalize plainly and stuckness

If Ralf didn't know it, I guess it deserves a name. This is the first I got,
feel free to improve.

That's the documentation I'd have wanted to see, and viceversa the lack of this
rule implied (to me) that this wouldn't work.

Thanks to Siddharth for pointing this out