When reverse engineering a lighting setup always apply the principal of Occam's Razor:
among competing hypotheses, the one with the fewest assumptions should be selected
Before introducing exotic lenses or beam splitting mirrors, let's examine the requirement for the incoming light to be perfectly aligned to the lens axis.
The bore is made of metal, and metal reflects light very well. If you shine a bright light into the barrel from any direction then it will bounce around inside illuminating the length of the barrel, removing the on axis restriction.
I don't know enough about firearms to say whether light could be introduced from the rear (presumably the same way the bullet enters the barrel), but again applying Occam's Razor, the simplest way is through the front!
The solution that seems simplest to me would be a multiple exposure with one bright light angled to illuminate inside the barrel, and a second designed to light up the front of the pistol. It's possible even that this was a single exposure, place a black plastic sheet touching the muzzle with small hole where the barrel is, hit it with a powerful strobe from and angle and whilst keeping the shutter open remove the sheet and let the ambient light burn in over several seconds. But a multiple exposure is simpler.