All theoretically possible risks with a reflector that sits directly on the inside of the glass window. Whether or not they are actually realistic is another matter.
A modified 6P LED drop-in is not "loose" in an 8NX in the sense of being completely unconstrained. Presumably, you've carefully filed down the centering ribs just enough to pass the drop-in through. If you do it right, the fit will be tight and the ribs will grip the drop-in's reflector body. Between the ribs and the outer spring, the drop-in will be held pretty well.
Granted, this is not as secure as in the X80 incandescent lamp case, where an overhanging lip is "pinched" in-between the ledge at the bezel window and the top of the collar. That makes the X80, bezel, and collar a single unit.
For the modified drop-in to move backwards and then spring forward again and hit the inside of the glass window, possibly breaking the window, you'd probably have to use a Kubotan thrust strike using the tailcap or accidentally drop the flashlight onto its tailcap. Momentum would cause the mod'ed drop-in to try to continue to move rearward, while the flashlight stops suddenly when it hits the target or ground. But a typical 6P drop-in is also very light weight, so it seems unlikely that it will have enough momentum to overcome the friction from the centering ribs and also compress the outer spring enough for the drop-in to lose contact with the window. The battery is far heavier, but it can't move rearward because it is blocked by a plastic ring at the bottom of the 8NX body, on which it sits.
Even if a gap did occur between the drop-in and the bezel window from a tailcap strike or dropping the light on the tailcap, how much force will be applied to the drop-in pushing it forward again? Just the outer spring, which doesn't seem to have a lot of "oomph" to accelerate the drop-in.
If you dropped the light onto its bezel, I don't see a problem in terms of the drop-in breaking the window. It's like shooting a shotgun. If you hold the buttstock in tight, recoil gives you a push with no bruising. If there is a gap, then the buttstock gives you a punch. Well, the drop-in will be in contact with the window, so all it will do is push on the glass.
If the window had a pre-existing flaw (a scratch), then you should take care of that regardless. But the drop-in reflector rim itself isn't going to do any scratching. It is soft aluminum, much softer than glass.
I suppose that repeated bezel or tailcap thrust strikes could repeatedly shift the drop-in and battery back and forth. For example, a bezel thrust could compress everything forward, and then the battery will move rearward again when you draw back your arm for the follow-up strike and because the outer spring pushes the battery back. The battery could move rearward faster than the spring and-or drop-in, resulting in a gap. Now we are essentially back to the tailcap thrust strike scenario from before. The cocking of the arm for the follow-up bezel thrust would have to move the drop-in rearward, resulting in loss of contact with the bezel window. Then the forward movement for the follow-up strike could slam the battery forward into the drop-in and the drop-in hits the bezel window.
All theoretically possible I suppose.
Of course, the filament on the X80 could have also broken in all of this abuse.
If this is still a concern, I would replace the glass window on the 8NX (the bezel has a threaded retaining ring) and drop in a Lexan window. Problem solved, though I'm not sure where to find a Lexan window for the 8NX. But that's a different problem.
If there is not gap to begin with, then forward-directed force against the bezel window is highly unlikely to do anything. This is the same theory when shooting shotguns and other heavy-recoiling guns. If you hold the buttstock in tight to the shoulder pocket, the recoil simply pushes you. If you are afraid of the gun and don't keep it in tight but instead leave a gap, then you get punched by the recoil.
Note also that SureFire clone bezels from Solarforce and G&P all are designed to have the reflector rim sit directly on the inside surface of the glass window. Those clone bezels do not use the same "ledge" design of the SureFire bezels. In the SF "ledge" design, the ledge is a protruding shelf. Thus, there is a ledge surface from both the outside and the inside of the bezel. The window sits on an O-ring which sits on the topside of the ledge and all of that is secured by a threaded retaining ring. A SureFire lamp or drop-in rim presses against the underside of the same ledge.
In the Solarforce/G&P bezel design, the inside diameter of the bezel is the same as the reflector rim diameter of a lamp or drop-in. The top opening of the bezel is wider, which forms the ledge. You drop the window directly onto the ledge, then an O-ring, and then the retaining ring.
I've not seen any complaints on CPF over this design in terms of causing broken windows due to lamp or drop-in impact to the glass.
Photos to follow later, showing these two bezel designs in pictures.