[roesel]

I have used the word "re-emit" in the sense of the Maxwell equations description of electromagnetic fields. You are right that the light definitely does not get absorbed and then re-emitted in the sense you mean. In such a case, everything you wrote would be correct. (I probably should have used a term like "radiate" and perhaps dipoles instead of electrons to be more clear.)

I was trying to describe the propagation of light in a material, where the optical field induces oscillations in the dipoles of the material, and these dipoles in turn excite the optical field. This happens constantly in every nanometer of the material, and it is difficult to experimentally separate the field into the "material" portion and the "vacuum" portion, because it exists as an everchanging mixture as long as there are dipoles around.

As for the "rails", the way I've had it explained to me is that in one direction of a polarizer, electrons are free to move, so they fully absorb the light polarized in that direction. In the perpendicular direction, they are bound, and the best they can do in reaction to a field is oscillate back and forth a tiny bit. These oscillations excite an optical field again and it propagates further until it finds another dipole to excite. I like to crudely imagine a polarizer as the grid of a nanoscopic egg slicer :D. Field oscillations will get absorbed along the metal wires, but in the perpendicular direction, it will just excite vibrations in the wires, which will radiate them out again, sort of like a guitar string.

Let me know if this was helpful, or if I've made a mistake somewhere :).