Merge branch 'debug/fem-order-adaptivity' into 'master'
use grid adaptivity in all orders and introduce order=3 This fixes #13. After all, it was just an issue of calling `adapt_grid()` with the correct arguments. I introduced `FEorder=3` as new supported order and removed `FEorder=0`. As reported to me at the Duner User Meeting, a higher than linear FEM order is desireable for DG problems. So I see no point in retaining a 0th order, i.e. piecewise constant functions. I tested adaptive grid refinement for all orders, all dimensions, and rectangular and simplex grids. Everything seems to work fine. Notably, the performance difference between the structured YaspGrid and the unstructured UG grid is quite strong for the third order - at least in Dune v2.4. See merge request !5
Showing with 23 additions and 22 deletions