Settings within VPVR and what they do. (For instance why resolution doesn't seem to change quality)
Settings within Steam VR (or why do we even need Steam VR)
How to check your performance, and not rely on re-projection.
Why does table look sharper the closer you get to it? What could help with sharpness from normal playing position?
Why Super Sampling in Steam is different than Oculus or within VPVR, or why you should use a combination?
To answer some of your questions more directly, they are not specific for VPVR but rather general VR questions.
Settings within VPVR and what they do. (For instance why resolution doesn't seem to change quality)
Each headset has it's own resolution, that's how the headset was designed according to hardware capabilities, you always render at the resolution that the headset is designed for, what you can do is supersample which increases or decreasess the resolution in software and then downscales/upscales it when rendering in hardware to increase or decrease image quality/performance but you don't "set" the resolution you get it from the hardware and the API.
Settings within Steam VR (or why do we even need Steam VR)
All VR software needs to use an API, SteamVR API is open, easy to use, good functionality, has a large range of support for headsets etc. If you used say Oculus API instead for example you would be limited to Oculus headsets.
How to check your performance, and not rely on re-projection.
In SteamVR settings you can enable a performance graph and choose to display it on the monitor, in the headset or both.
Why does table look sharper the closer you get to it? What could help with sharpness from normal playing position?
Limitations in current VR hardware, the resolution and pixel density are limiting, a more high-end headset gives better results, for example Valve Index.
Why Super Sampling in Steam is different than Oculus or within VPVR, or why you should use a combination?
It's not different, it produces the same result, it increases the resolution in software before downscaling/upscaling it to hardware for either increasing quality or performance. You should always only have supersampling in one place not multiple because it's additive, VPVR will ask SteamVR API for the resolution of the headset, if you have supersampling enabled in SteamVR you will get the supersampled resolution, i.e. 1000x1000 x1.5 = 1500x1500, if you then set supersampling inside VPVR then you will supersample the already supersampled resolution inside VPVR as well 1500x1500 x1.5 = 2250x2250, this will impact your performance greatly and probably not produce the results you were expecting. Supersampling has diminishing returns, after a certain rate you will not notice the effect greatly but it will have a big impact on performance since you need to render more pixels, there are a lot of discussions and debates about what to set it to, some claim they don't see a difference after 1.5, some claim they see a big difference up to 2.5 however there is a huge performance difference between 1.5 and 2.5 (more than twice the resolution that you have to render in software).