Good luck in your mission. There's no real consensus on mesh density, at least none that I've seen. You might as well do what works best for you. As far as Sub-D, the program doesn't save the increased geometry in the obj, it's only used internally in that specific program. Each program does it's own form of subdivision. So in Daz Studio, once you subdivide something, it will use more system RAM and more VRAM for rendering, but saving that model as obj will just be the original model at base resolution.
In DS, if you view a model in Lit Wireframe mode, you can see the base geometry as dark gray lines, and the subdivided geometry as light gray lines. Aside from that, it doesn't really make a difference, the object just looks smoother.
1. When you load a product in DS, it may or may not have subdivision already enabled, depending on whether the artist wanted to enable it. If it doesn't have Sub-D enabled, you can enable it through the menu (as you already know) but your results will be hit-or-miss. If a model isn't properly made for Sub-D, it's going to look horrible.
2. Like I mentioned before, it adds geometry, but only for that specific program, you can't export it.
Here's a good explanation of how Sub-D behaves in DS https://www.daz3d.com/forums/discussion/comment/2583171/#Comment_2583171 . Basically, there are 2 settings, RenderSubD Level and SubDivision Level. RenderSubD Level only affects Iray renders. SubDivision Level affects Iray preview, viewport window and external renderers (Luxrender, Octane, etc).
I haven't even tried looking at triangles yet. Hope this helped.