That's awesome! And the map itself looks very nice! 
Also it's very nice
I'm currently taking a break after the 0.56.2 release but when I'll be back I'll host it on my server! 
I was surprised you used CRN for lightmaps, but it looks like they are so high res that they avoid the usual artifacts we can expect from CRN lightmaps and other lossy formats… We may investigate the usage of the -fill q3map2 option meant to workaround jpeg lightmaps, that is expected to be useful with lossy formats in general.
Actually I find this very interesting, CRN reduces VRAM usage but produces artifacts, high res lightmap increases VRAM usage, but if high res lightmap also reduce artifacts, then we may find a sweet spot where CRN lightmaps allow both increased resolution while preventing artifacts and preventing crazy VRAM usage. We cannot remove the computational cost of high res lightmaps though… And the DPK size can be felt. 
I love how detailed are the lightmaps. For the cast shadows of grates in front of lights, I have not checked the shader source but I have the feeling they behave as if light subdivision was low (high subdivision size, like the default one), basically, like if the whole neon was a single point. Reducing subdivision size (increasing subdivision) would remove the “all shadows come from one origin point” feeling but also blurry the shadows (which is in fact more realist, when the light comes from a large light like a neon). That's also a matter of taste though.
Now I want to investigate that high res CRN lightmap thing. I'll see if that fits the metro rebuild I'm working on.
What was the approximate compute time for the CTCS lightmaps on your end?
Out of topic, but I see that's (at least) you're third account on the forums. If you need to recover an account I can help you with that.
