The "buffer" reduces decision complexity even more because people treat them like train blocks. The only annoyance I have is when people actually break-and-check at these points even though its better to roll the car slowly trough to save the people right behind from brake checking entire queues.
To eliminate this you could turn the buffer into a whole extra lane with room for say 5 cars to queue, but this would compromise on the nice feature where the partially turned car gets to completely turn and have great vision of the cycle lanes in both directions.
It's an interesting article, but from a systems design perspective I'd be much more interested in how they handle a change in requirements like "there are now five times more cars turning left here than the intersection was designed for".
To an extent, it's a self-solving problem. If you have great non-car transport options and an increase in traffic makes car driving less appealing, then more people will use those non-car transport options rather than joining the queue.
The problem is that you may not have the room for it. The US might often have more room to retro-fit bike lanes, due to their roads be generally pretty wide. European cities, like Copenhagen have a massive issue as more and more people get things like cargo bikes and electric bikes. The bike lanes needs to be expanded to accommodate them, but there's no room. You'd have to remove cars from large parts of the city, which sounds great, except you do need to have the option to drive, either due to distances, public transport or deliveries. You can't do parking and have people walk, because there's also no room for parking.
For some cities I also don't see bike lanes as solving to much. Some cities, again often in the US have a huge area and millions of people. Distances in cities like Houston, New York, Los Angeles or Atlanta are just insane, taking up enough space to cover half of a small European nation.
Without the buffer, a single car wanting to turn that way when there is a cycle in the lane would block traffic, unless of course the car takes priority and just expects the cyclist to deal with them cutting in front (which is my experience too often at junctions with or without cycle lanes…). In either case, with or without this design, the car slowing down to turn is going to create some back pressure if the road is busy, there is no avoiding that and this design might even actually slightly reduce that issue.
Looking at the picture I assume that most vehicles are going to be going straight on, and when someone is turning the only extra delay is when their need to turn coincides with there being cyclists or pedestrians in range of crossing, so it is likely that none of this back pressure is a problem the vast majority of the time.
If it becomes structural, say the neighbourhood becomes larger and substantially more cars will go there now, then the intersection will be redesigned. Money isn't infinite of course, but this sort of thing is a big part of planning new development.
I don't know any concrete example, but since road engineers have been using queueing theory, originally invented for telecommunication networks, for more than 70 years, I would be surprised if models and tools designed for one use case had not been reused for the other.