I've written before about the importance of keeping track of what comes from Okrand, but I think there's a useful difference between, say, Okrandian canon (comes from Okrand) and Okrandian Klingon canon (Star Trek canon which uses Okrand's Klingon). 

Okrand himself specifies the rules for generating noun-noun constructions and compound nouns, for example, in TKD. Since {poH qut} has been used in Star Trek now, and its meaning is fairly unambiguous from the components, I see no reason not to accept this as the "canon" way to say "time crystals".

There is, as you say, a difference between Okrandian canon and Star Trek canon. If poH qut appears on a Star Trek show but wasn't written or approved by Okrand, it is Star Trek canon, not Okrandian canon. On this list, canon means, first and foremost, Okrandian canon.

(TKD even allows such words to be written without a space, like *{poHqut}.) I'm much more likely to use {poH qut} than {jolvoy'} (since Discovery is currently on-air), and also be likelier to be understood. {poH qut} is Okrandian Klingon canon, even if it isn't Okrandian canon.

"Okrandian Klingon canon" isn't a thing. If someone uses grammatically correct Klingon on Star Trek, that doesn't make it closer to Okrandian canon than if someone gets it wrong.

While most episodes of Star Trek: The Next Generation feature Klingon that the writers just made up, some of them give us Klingon that was obviously taken from The Klingon Dictionary. We don't call these Okrandian canon. Okrand usually backfit most of their errors into the language by giving us new canon, but the original lines still aren't Okrandian canon.

I can virtually guarantee that if someone were to show that line to Okrand with the explanation that L'Rell spoke it on Discovery, he'll confirm that it is correct. Again, it's fine to distinguish between "it comes from Okrand" and "it was approved by Okrand" (as some of K.R.A. DeCandido's words are), and it's fine to wait until he does confirm it and not just assume he would (though he has a lot of things to do, and I've never known him to contradict on-screen Star Trek canon). But I think there's also a danger, in absolutely refusing to interpret a word in any way except as illustrated by an existing canon example, of being more Klingon than Kahless (I think the Earth expression is "being more Catholic than the Pope"). 

I think you've constructed a straw man argument here. Nobody has demanded rigid adherence to the exact words Okrand gave us regarding SIch. Some have probed the limits of what he said, and I, for one, have said "the jury is out." I'm not going to complain if someone uses SIch in a way that matches what L'Rell says, and I can't imagine anyone else will either. We simply have what Okrand has said about SIch on the one hand ("canon") and what has appeared on Star Trek on the other. They're not the same thing. When trying to draw conclusions about what bits of Klingon that appear on Star Trek, we don't get to pick and choose which bits count as canon and which don't.

The definition of canon is not "what we imagine Okrand would say." At the same time, when speaking or writing Klingon, one is not obligated to obey all conclusions drawn from canon. You just have to be willing to agree that what you said is not supported, or not clearly supported, by canon.

As an example: for years this list used the word pabpo' to refer to grammarians. Not canonical. But no one ever claimed it was canonical.

-- 
SuStel
http://trimboli.name