Projectors have limitations to begin with. And this idea takes all of those limitations, and adds even more inefficiency to them.
A major hurdle in projection that still remains today is brightness.
A projection screen that offers 50% optical transparency throws away at least 50% of the light energy that hits it: Of all the light being directed towards that screen to produce an image, no more than half of it can be reflected.
And furthermore, a screen with 50% transparency doesn't look very transparent at all. A hypothetical frosted screen that is suitable for projection with 50% reflection and 50% transmission might not look even a tiny bit like a window.
In short-term visual effects for stagecraft and amusements, that's not a big deal. "Oh, we lose fifty or sixty or ninety percent of the light we project for this effect? No worries; we'll just use a bigger, brighter projector until the effect works. Call Andy over at the rental house and see what he can get for us."
But in long-term use, that kind of built-in inefficiency is mostly a non-starter: It's inefficient to over-build the equipment in this way, and it's inefficient to burn (waste) the energy to keep it going.
Maybe with some massive improvements in material science we can have a semi-transparent screen that concurrently reflects and transmits light over-unity, but we're not there yet. But to hazard a guess, it's probably easier to accelerate a mass beyond the speed of light than to pull off that particular trick.
(Meanwhile, see-through/transparent LED-based displays with decent-ish optical transmission? Whether good or bad, that technology seems to be right around the corner.)
So TV < laser < bulb. The short throw design doesn't seem to change the numbers much.
A projector that is shining upon an opaque white surface is a whole different thing than one that is shining upon a surface that is transparent.