Merge branch 'feature/fix-typo' into 'master'

See merge request [extensions/dune-codegen!336]

  [extensions/dune-codegen!336]: Noneextensions/dune-codegen/merge_requests/336

Merge branch 'feature/cache-basis-in-constructor-blockstructured' into 'master'

See merge request [extensions/dune-codegen!335]

  [extensions/dune-codegen!335]: Noneextensions/dune-codegen/merge_requests/335

Merge branch 'feature/blockstructured-remove-clutter' into 'master'

See merge request [extensions/dune-codegen!323]

  [extensions/dune-codegen!323]: Noneextensions/dune-codegen/merge_requests/323

Merge branch 'feature/fenicsesque-finite-elements' into 'master'

ref:extensions/dune-codegen TODOs:

-   [x] Register all finite elements from PDELab into UFL
-   [ ] Adapt driver generation so that they are actually used
-   [ ] Adapt examples to be more precise about the basis they are using

See merge request [extensions/dune-codegen!322]

  [extensions/dune-codegen!322]: gitlab.dune-project.org/extensions/dune-codegen/merge_requests/322

Merge branch 'feature/sumfact-loop-reordering' into 'master'

ref:extensions/dune-codegen Performance transformation through loop nest
reordering. There are two ways to reorder loops in a tensor contraction:

1.  Directly accumulate in output variable after setting to zero
2.  Accumulating in a large enough temporary

This merge request implements these ways of loop reordering and the
possibility to create an autotune target directly from the loopy kernel.

See merge request [extensions/dune-codegen!305]

  [extensions/dune-codegen!305]: gitlab.dune-project.org/extensions/dune-codegen/merge_requests/305

Run the function call over multiple data sets.

Replace placeholders in template file with actual code. This makes it a lot
easier to restructure the benchmark.

This rewrite was necessary for the cases where the instruction contains a
reduction but the expression itself is not a reduction. Besides fixing this bug
it greatly reduces code duplication and increases code quality.

This just reverts to the old behaviour. The new implementation of loop
reordering doesn't need that anymore.

Merge branch 'feature/cmake-doc' into 'master'

See merge request [extensions/dune-codegen!324]

  [extensions/dune-codegen!324]: Noneextensions/dune-codegen/merge_requests/324

Still missing: Handling of reduction with haddsubst!

- Code cleanup
- Use loopys duplicate_inames
- 2D, 3D volumes and faces should all be working

This can be used to distinguish between kernels with the same name but
different transformations.