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

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

* Move coefficient computation into RuinaNonlinearity
* Allow some friction data to vary in space

As a consequence, the nonlinearity is only assembled once

Also fix the order a-mu-V0-stress-b-state-L everywhere