Sure, why not? But it contains numerous practical problems. You still have cars crossing streams with cyclists. You will not be able to enforce the 25 mph designation, particularly since you'd have to discriminate between the speed of the inner lanes vs. the speed of the outer lanes. And this takes up much more space than a simple bike lane, with or without an extra shoulder.

So, you've made it more expensive to impliment, more complicated, and there really are no advantages. Bike lanes already allow merges by cars into them at intersections. Large volume crossings are easily handled by RTOLs, which is essentially identical to a merge into the bike lane. Low volume crossings are irrelevent in my experience.

All you've done is create a dedicated "get out of traffic jam free" lane for motorists to shoot through and have changed nothing for cyclists over a bike lane. I agree with, perhaps, widening the bike lane. This deals with the visibility aspect and the "not a real lane" aspect. But at the same time, my experience here, vs. the experience of others in regions where cars are allowed to merge into the bike lane, show that the limited access feature of the bike lane is essential to its effectiveness. It implicitly teaches motorists to regard the bike lane space, making it harder for a cyclist in the bike lane to be overlooked and rendering the "inadvertent drift" argument moot. After all, if you have to avoid something, you cannot help noticing it. If you (the driver) are freely allowed access for turning maneuvers, you can get through life without having to recognize the presence of the bike lane.