If the tech exists, subtle is going to be a design choice.
The fact of the matter is that it's kind of pointless to worry about it from the standpoint of modder difficulty; Alex is quite unlikely to want to change the core aesthetic this late in production, guys. This isn't like WTF, where the coder's still trying to make his slow voxel thing look decent at crawling framerates, lol.
From a practical standpoint, it'd work well for TCs but practically nothing else; I absolutely agree that it'd be silly to demand that David spend the next couple of months retro-fitting everything with normalmaps, lol.
If Alex allows hooking into the engine at the point sprites are drawn, somebody will want to do normalmaps; if they're done right, they'll look a lot different than Vanilla sprites. Not necessarily "better", but they'll certainly look more realistic.
In terms of production times and all that... this evening, I looked at the practical problems of production workflow, and it's not that bad... if you're a professional 3D artist. You can just output a high-poly sculpt to a normal map, like you'd typically do for tiles, render a diffuse map and build the material map and you're done. Most of the time spent would be on the sculpt; pixel-art detailing would be less important, because the normal map would largely deal with lighting and we could use an AO channel in the material map to handle the rest of it.
Nothing else about the rendering process would change much; the damage overlays are designed to work with anything they're rendered over and would probably still look decent, although if that system could be overridden and current "heat" levels were readable by mod code, even more attractive possibilities present themselves.
The only big fly in the ointment that I found is that resolution on the normal maps would probably need to be 2X sprite resolutions, largely to produce enough slope in all circumstances. I don't think sampling on the GPU can quite get it done at the scales required without the occasional issue with shimmer or weird height reads, but that's all conjecture atm, I just did some tests of some things to see what looks good.
All that said... meh. Unless Alex wants to license the code as an engine and farm it out for other people wanting to build similar games... I can't really imagine anybody going to that much effort, tbh, except on the smallest scales.
There are maybe 5-10 people around here who might be able to do it, art-wise, but practically nobody would be willing to sink the time in, for free. The amount of time investment required at that point it serious enough that I'd be very surprised if somebody even found the time to do all the sculpts just to enhance the Vanilla stuff properly, tbh. I sure can't see Alex and David getting all excited about having to re-master all that art themselves.