Fixed problems in computation of SymmetricComponent.integrate found by new unit tests...

Fixed problems in computation of SymmetricComponent.integrate found by new unit tests.  There were two problems with the previous code.  First due to rounding errors, it was possible during the last iteration, the right end point (x+l) was beyond the overall length of the component.  This resulted in a zero 0 being used for r2 throwing off the overall computation of volume.  The second problem was for hollow components.  The test for when to switch from using filled to hollow was incorrect because it was comparing r to thickness.  However, thickness is normal to the surface of the component so this test overestimated the total volume of the component.

After making these computation changes, the tests for the density computation in  NoseConePresetTests and TransitionPresetTests could be tightened.

git-svn-id: https://openrocket.svn.sourceforge.net/svnroot/openrocket/trunk@590 180e2498-e6e9-4542-8430-84ac67f01cd8