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

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.

The SumfactKernel object stores a tuple of transformations that will be
registered in the realization and applied in the localoperator. These
transformations can change the name of the sum factorization kernel
function. This way it should be possible to use autotuning for picking
transformations.

Decision about loop order should be done with autotuning in the end.

This commit includes some ugly code that needs to be cleaned up!