Wait just a moment. You asserted a principle. We do not see evidence that you demonstrated it. About the only empirical aspect of that picture is that the picture itself exists.
I suppose it might depend a bit on the coherence of your light source and the reflectivity and distance of the clouds, but when I actually tried the diagram you post, with the light source where you show it, it left the table entirely unlit, and the paper ceiling showed a fairly sharp delineation between light and dark, corresponding to the shape of the table edge. If the light source continues to sink further below the table, the lighted area moves, of course, steadily toward the edge. Since it is reasonable to assume that a flat earth's atmosphere is also flat, and does not extend into outer space, it also has an edge, so the light diminishes into a smaller and smaller slice, visible equally and simultaneously to all upon the table, until it abruptly ceases. The apparently temporary half-answer that the sun might not go below the surface of the table requires a pretty elaborate bit of mumbo-jumbo theory to allow a sunset and sunrise.
It is hard to figure out a flat earth scenario in which there is not either instant night over all the earth at once, or no night at all. You may be quite able to function with a flat earth assumption if you don't think about it at all, and just count it a black-box miracle, but pretty hard not to consider a round earth empirically obvious if you do.
Granted, there may be ways to make it work, and maybe it's still possible somewhere to take Phoebus's golden chariot as nonfiction, but I do not think that that theory passes empirical muster.
e.t.a. on the issue raised by Reformed Offlian, I think you're reading it wrong. The phenomenon of the lit clouds could be observed by anyone on the flat earth who can see the distance. But wherever you are when you see it, it would always have to be at the same time, in the same place, moving in the same way toward the sunset edge of the table.
And a further e.t.a. I think there is some backwards thinking here on another front. In this issue you need not come up with a proof of what is. Only of what isn't. It may be that a round world is the only reasonable alternative to a flat one, but you don't have to come up with arguments for roundness. In this case, all you need to do is show convincingly why the world cannot be flat. If the flat earth theory cannot survive the sunset, it's over, even if you then want to go on to mobius multiverses, wormholes and time cubes.
