At least my emulsions (BrI) were somewhat "transparent" in the sense that they might work in a multilayer setup with limited sharpness and limited speed... I mean, of course they are "milky" or diffuse, but if you'd place a newspaper in a close contact with the emulsion, you might be able to read the text through the emulsion. So when the distance gets shorter, "translucent" changes to "transparent".
Of course, the wavelength which the grains are sensitive to is absorbed by the grains, but what happens to other wavelengths? They do not go through the grain which would be optimum, right? Instead, they are reflected by the grains and get scattered. So, to make the top (blue) layer transparent enough, it needs to be "inefficient" so that it passes part of the light through it by having fewer grains than would be optimum for single-layer? Then, the blue wavelengths of the "pass-through light" is absorbed by the filter layer. I guess there are more optimum ways because what I propose wastes light quite a bit?