1. Narrowing ranges
If I understand this principle when we move from one decision to the next decision we should only include the part of our range that took the actions. There's nothing wrong with starting ABCD already before the flop and the continuing with (for example) the B range when we have called pre-flop and expand that into ABCD ranges on the flop. For example in Stacks' example the 88-KK hands there were B range on the flop should not be in any of the ranges on the turn since we did not call.

This is true, I actually didn't notice.

When we start balancing ranges that will invalidate the point, but for the exercise and clarity I think it's best to stick to polar ranges.

2. Full ring forum?
I think it's safe to say that the ABCD theorem is applicable also to short-handed play - and I think it's an absolutely terrific tool. Should we post 6max hands in here also? Or could ABCD theorem coverage be moved to the generic NL Strategies forum?

Yes, please do.

3. Modifying ABCD ranges based on reads
The easiest example of this thought is pre-flop (as evidenced in the 3b/4b thread in the Strategies forum). For example if an opponent folds to a 3bet 99% of the time (only calling/raising AA/KK presumably) we could beef up our C and B ranges against this opponent considerably. Similarly if we know an opponent never C-bets missed flops we can beef up our B range in position, bet when checked to and take down the pot etc etc.

Instant Aces (I'm probably not particularly qualified to comment):
Maybe KdJd / KdTd / QdTd in A range on the flop if any of them call PF in position against this opponent (and KT in B) - otherwise I like the flop ranges and note that QQ-AA and AK are 3betting hands pre-flop and thus would not have been called.
On the turn I agree that all trips are A range, but I would hesitate to put 99 in A for value. While it's true a 7x hand might call a bet, which would make 99 betting be for value, I think it's easier to extract value from 7x type hands on the river. I would probably put 99 in B. I agree with C range.
Interestingly, the check/call on the flop means that on the river we have no full houses or flushes in our range. If our opponent is perceptive and has kept this kind of track of our ranges (ok he's a monkey and probably hasn't) he'd know that a bet can't mean a full house or a flush. This also underlines that it might be beneficial on the flop to put some of the 33,77,TT and flush draws in the B range a percentage of the time. (I believe this is what is called balancing the range). On the actual range (assumption being an absence of balancing) I would leave out 88, 99, 87 as checked on the turn and move the trips with lesser kickers into B (JTs, T9s - maybe even QTs).
Of course there needs to be enough hands in the A range to justify an amount of hands in the C range (playing for folds) so I am not sure demoting the lesser trips is correct to allow for some hands in the C range. But on the other hand trips is a hand with relatively middling hand strength at this point (full houses and flushes) so I don't necessarily want to bet a great amount of them. Great turn bet sizing btw, and very perceptive to include hands with blockers in the C range on the river.
Further to the river I think I'd like to discuss which hands will call and which hands will fold here. I think the opponent can show up on the river with AA, KK, QQ, JJ, 99, 88, 87 type hands that will fold (or consider a fold) if we shove. TT, 77, 33, T7 and flushes will all call our shoves. This suggests to me that we can't really value shove at all with any trip hands. Maybe we don't have an A range at all. I don't think I'd have a problem with all our trips in B and if we think he'll fold AA to a shove just expand the C range to all overcards with a diamond. It seems exploitative to me to have a bluff range without having a value range, but I guess we can argue that some flopped full houses and flush draws were checked for balance and those are the ones we are representing.