This is generally how I see it going, but note this is speculation.
1) When a city first is formed, it is a big unhappiness magnet. Its growth is quick, and with little infrastructure it produces few yields. However, this is partially controlled by the population cap on unhappiness.
2) As the city starts to grow into its own unhappiness drops. The city is working its best tiles, its buildings are giving it very nice bonuses. In fact, I would wager that its unhappiness is lower than you might expect, because its producing so much compared to its population.
3) As the city starts to grow large, the bonuses start to decline. Specialist slots are filling up, the terrain tiles aren't as good as the ones the city is already working, and the buildings just aren't bringing that extra boost of resources like before. However, because the city was overproducing yields before, its just now "producing well". Unhappiness is more than before but manageable and reasonable.
4) As the grows very large, its extra bonus per population is now minimal. Specialist slots are full, terrain is mostly tapped. Buildings add a few yields, but compared to the large pile of yields already there its a drop in the bucket. And so the yield per population drops and drops. This is the danger zone, where population is no longer providing the yields to self-sustain, and there is no way to counteract it except stopping growth entirely.
So I think number 4 is the issue, and probably because (I image) the unhappiness system is somewhat linear in its approach. Its yield per population. That works well for phases 1-3, but then starts to fail out in phase 4 where getting new pop doesn't really translate into getting new permanent yields.
Assuming there is a true problem here, I can see a few solutions:
1) Add a population counter adjustment to needs. Similar to the capital adjustment, cities above X would get a "high pop city" adjustment which would lower the needs a bit. In effect we are adjusting the model to account for the fact that yield per pop starts off linear (or maybe exponential) but then tapers off at the higher populations.
2) Add a happiness bonus to some building for very high pop numbers, or a new building entirely. This is similar to some of the old sewer system equivalents, where you would build buildings specially to counteract the effects of high pop cities. This would is much more blatant and obvious correction, but it would give the player a very direct way to address the problem.
3) Now adjusting the luxury system may help, but I feel that either you make the system too easy for Tall, or make it that Wide can get away with loads of big cities and Tall still suffers. I may be wrong on that, I just feel that luxuries are a bit variable and not the way to address it directly.
So my thoughts would be on number 1.....again, assuming there is a problem.
An excellent point. How many exceptional tiles are there for any of us to plant a city down? 5-6? Once those are worked, probably around 8-10

after the requisite food tiles, then every additional worker is working marginal tiles. Your yields per

can only go down from there.
This is the problem with strategies that try to leverage tall cities, after 20

, the infrastructure and yields potential of cities simply dries up.
You will always want 1-2 large cities, even if they become drains on happiness, because the game rewards large cities in other ways. Only large cities can leverage the

required to win wonder races. Large cities are necessary for winning wars, and any short to medium term goals, like blitzing diplo units, military units and settlers, etc. require that you have a city with

to spare. This means that your smaller cities are propping up your population centers, but the revolt mechanic seems to undermine this. If you get really unhappy, the revolt system cuts away at all the cities which were keeping you solvent, starting with the puppets, which are often the smallest drain on your happiness.
I'm entering Industrial now with a new game with Iroquois. I am at +32

, but I am standing at the precipice. My capital has 21

, and is generating 5

. Iroquois' playstyle encourages many small cities which can still be highly productive, so they are perfect for this patch. I am under the impression that the game mechanics, as they are now, heavily favor this kind of civ.
Tradition civs are so last season; Progress is king.
RIP: Arabia, India, China, Egypt, Korea, Maya, Venice, and 4UC Babylon
Long live our ascendant masters: Iroquois, Songhai, Carthage, Shoshone, Spain, Polynesia
make it harder to get there but not have it be punishing. This way he'd have no real unhappiness problems, but only about 37-45 pop at that point because it'd cost more.
Making growth past 20

cost a lot more food is certainly an attractive possibility. It doesn’t directly punish any 1 civ, and makes the civs that can break 30

in a game truly exceptional.
If the curve for

city growth was made steeper, that would mitigate the ultra-tall problems, and combined with a boost to the efficacy of luxuries could be all that’s required.
Returning to @Stalker0’s three solutions, I also lean towards option 1. However, the most immediate problem with happiness right now is that luxuries are weak. We can re-assess the happiness system when luxuries are working as intended.