BUG: Fixed error in area-weighted normal
When the closest point on a surface to a test point is on an edge, the area-weighted normal is the average of the normals of the faces adjacent to the edge. The contribution from one of the faces was being added twice, producing the incorrect area-weighted normal. This resulted in the wrong sign for the signed distance in some cases. This patch fixes the error. Changed TestImplicitPolyDataDistance.cxx to fail prior to this change and pass after the change. The previous test exercised vtkImplicityPolyDataDistance but did not check the results. The new test checks against a baseline.
Showing with 124 additions and 70 deletions