You can save the paths as usual, save selections, layers, etc. Same with smart objects, save the layer (merge a copy of the layer into the 3D model, so you can keep the original layer).
Cameras also have custom presets you can save to match a layer position, so if you have to adjust a shape you can always go back to the same camera / layer view and do it there without having to do it all over again.
As for the alpha for the decal thing, using levels adjustment on the final skin can turn it into a black and white alpha channel, or as I've mentioned before, saving a camera for each layer will get you the decals in the exact same position to do a black and white mask.
With all this, you save yourself the trouble of dealing with the UV, is more of a "what you see is what you get" workflow.
So you can still draw vector paths on the 3D model (projected from a defined viewpoint)? And apply vector shapes/smart objects on 3D (projected from a defined viewpoint)? If those aren't possible, then the workflow becomes convoluted. That's still better than I was able to find when I last looked (I Googled "vector shapes Photoshop 3D painting" and similar for an hour the other day and came up with absolutely nothing). There's still the issue of projecting a flat object onto a complex 3D shape, though - if you want a logo to wrap up and over the fender, you're still going to get it skewed towards the ends of it, unless it can somehow intelligently wrap it around the corner. Is that possible? Otherwise you
still have to paste two copies of it, from two different views, and manually knit them together around the corner - and then you have to knit them with rasters.
Levels adjustment is not sufficient for alpha channels - that will make changes based on the colour of what is being used, plus you run into anti-aliasing issues where it'll pick up some of the hue, but not all of it, etc. The colour of an item has absolutely nothing to do with the alpha channel - a red logo and a blue logo and a yellow logo may all need to be the same shade of grey in the alpha, while the black and white paint may need to be the same light grey, and the orange may need to be black in the alpha. You can't do that with levels. Having to go back to a particular view and try to pick colours out of a logo to make a proper alpha is also convoluted. The reason I use vector shapes and vector smart objects is so that I can
very easily change them to the desired greyscale (or other colours, like for the _map texture) with a couple clicks
at most, with 100% accuracy and absolutely no anti-aliasing issues. If I know what colours I need things, the alpha channel and _map textures take me a minute combined to produce, whereas if I had to screw around with levels, colour replacer or saved selections, I'd spend hours. Working in vector makes changes
far easier than working in raster, even after you factor in the minor nuisance of alignment across a UV gap. I'd rather spend a few minutes aligning a graphic across the 2D UV gap than spend an hour trying to pick selections without missing those couple anti-aliased pixels so I can make my alpha channel.