My understanding, which could be wrong, is that it doesn't 'bounce around' as much as it is absorbed and re-emitted. This process isn't instantaneous and accounts for the net propagation delay through the medium.
Both ideas are incorrect. Bouncing around would be random and the outgoing light would be scattered (as in matte reflections) in all directions. Absorbtion simply destroys the original light. If photons are mostly absorbed, we call that “a wall” or ”a fog”, not “a transparent medium”, which to be transparent shouldn’t have electrons whose energy level differences correspond to a photon’s wave length. The transparency and slowdown effect is a complex interaction between fields in the medium, which I honestly don’t have enough courage to retell.
Sort of. Refraction it better looked at from a wave perspective than a particle one, IMO. The wavefront expands at the speed of light, but the wave bounces off of the medium it's traveling through and self interacts in a way that means that the highest probability area moves slower than C despite the edge of the wavefront still moving at C. That smearing is also why the wavelength changes upon refraction. (And obviously leaving concepts like time dilation off the table for simplicity's sake).
