Use separate bolt density

[nepfaff]

This doesn't really make a difference, but it seems like the right thing to do. This is based on the linked bolts in the README.

Before:

python3 GenerateObjectConfiguration.py -b 0 14 -s 0 1 2 3 4 -o test.urdf --print
test.urdf
Total Mass:      1.1718997410246232
Center of Mass:  [0.0000000000 0.0409352866 0.0250000000]
Inertia Tensor wrt. origin, about COM:
 [[0.0171855735 0.0000000000 0.0000000000]
 [0.0000000000 0.0040224473 -0.0023986026]
 [0.0000000000 -0.0023986026 0.0179134108]]

After:

python3 GenerateObjectConfiguration.py -b 0 14 -s 0 1 2 3 4 -o test.urdf --print
test.urdf
Total Mass:      1.1680782668800922
Center of Mass:  [0.0000000000 0.0410692101 0.0250000000]
Inertia Tensor wrt. origin, about COM:
 [[0.0171004315 0.0000000000 0.0000000000]
 [0.0000000000 0.0040168639 -0.0023986026]
 [0.0000000000 -0.0023986026 0.0178338314]]

[PhilNad]

Indeed, since the bolt is modeled as a cylinder but is not a perfect cylinder, its mass will be slighter less than was would be obtained from the steel mass density and the cylindrical shape.

Also, currently, the head and nut of the bolt are not taken into account (as you can see their impact is very small). I wonder what would be the best approach to model the bolts without spending too much effort on very small details.