This works because interpolateP0ToP1 is in dune-fufem now

* Use PointTractionBoundaryAssembler: The normal stress is properly calculated here
* Incorporate weights into normal stress

See if our alternating Gauss-Seidel scheme can really minimise distances

(1) Determine distance from patch properly (also in general polygon cases)
(2) Move refine()
(3) Rewrite the logic of the patch (allow it to be a polygon)

This allows us to return a GlobalFriction and thus hide some information

Also changes the [Output]

Solve until the binary comparison terminates
Re-enable fastQuadratic