The shells are not used in making peanut butter. You would need to shell these peanuts (and remove the papery skins from the individual peanuts) before grinding them to make the peanut butter.
It doesn't matter whether they are roasted in the shell or not--what matters is that they are roasted, to give the deeper, richer flavor.
I have to assume your peanut vendor did not mislead you, and therefore the peanuts you bought are already roasted. You would just shell them and grind them. Note: if they are roasted, they will be somewhat crunchy, with a toasty complex aroma, and light medium beige to brown color, whereas raw (or "green") peanuts will be closer to very pale yellowish beige, like unfinished lumber.
Due to the large number of peanuts that may be used in making a reasonable quality of peanut butter, and the inconvenience of shelling those peanuts, one might choose to purchase already shelled, roasted, preferably unsalted peanuts for the purpose.
Edit: in response to the second question, I defer to the mighty Alton Brown whose recipe indicates to roast peanuts at 350 F for 30-35 minutes. Since he doesn't give us a test or indicator to know when they are done, and a careful and meticulous recipe writer such as Brown would not omit this if one exists, I infer no good test for doneness of the in-shell peanuts exists.
However, since the individual peanuts within the shells are very uniform in size, the time and temperature guideline is likely to be quite effective.
You should almost certainly be smelling roasty peanut aromas as the peanuts close in to being done, but there is no way to describe that. Its like knowing when a cake is done by the aroma--its certainly possible, but you have to have experience with the recipe to know.
While the grinding of peanuts produces a paste we call peanut butter, a food processor may be slightly better equipped to handle them then a coffee grinder, especially if you want to blend coffee with it again.
Peppercorns are harder and they can be done in a coffee grinder, but peanuts would likely result in a mess that would not easily be cleaned, and somewhat chunky peanut butter.
edit:
As stated in the comments below you may get some good headway with a blender. I had a crappy blender at home, but peanuts may be easier to grind than thick smoothies. But again, cleaning may take some time afterwords. But blender would probably be more effective than a coffee grinder.
Peanut powder can be made by freezing the peanuts I believe, very cold. A part of the modern cuisine. I think you would need liquid nitrogen for it. Source from Tech stuff podcast, title of the podcast was Molecularly Gastronomical.
Best Answer
There's definitely some rounding going on because the peanut butter has 100.1g of nutrients per 100g of product. This isn't enough to explain the discrepancy. Adding up the nutrients on the roasted peanuts gives 95.4g. I think we can assume the other 4.6% is water. So perhaps more water has been driven off the peanut butter.
What I think is more likely (though could be additional) is that there's significantly less fat in the peanut butter (48.1% vs 51.7%). While this could be down to different varieties of peanut, I suspect some fat is removed in the processing, perhaps to avoid an oil slick on the surface.
Most likely of all is just different source data. Different peanuts may have been used in the calculations -- whether that is consistent with the actual ingredients used is another matter. In fact if you look at nutrition information for raw peanuts online, you'll get a range of values. There's nothing special about peanuts, the same is true for bananas. I'm sure if you look at the scientific literature on any foodstuff, you'll find a range of values published, reflecting natural variation as well as measurement variabilility -- this will then propagate to the values published to the consumer.