Fix mortar assembly

1. In dualmortarcoupling.cc: Currently while assembly on a boundary intersection the mortar FE basis functions are considered whose associated vertex belongs to the inside element and lies on the restricted boundary. Unfortunately in some three dimensional grids it may happen that all four vertices of a tetrahedron lie on the (restricted) boundary, which leads to wrong values (since the mortar FE basis function associated with a vertex which lies opposite to the current boundary segment/intersection does not vanish on the intersection.) I added a further condition, which checks whether the associated vertex of a mortar FE basis function actually lies on the boundary segment (and not opposite to it). Maybe there is a smarter way to fix this.

2. Fixed a typing error in nbodyassembler.cc.