[!328] Performance measuring with Intel SDE

Merge branch 'feature/sde-performance-measuring' into 'master'

ref:extensions/dune-codegen This allows to use [Intel SDE] for performance
measurements. SDE is capable of counting exactly the executed FLOP and
transferred Byte of a program. The output is quite excessive, but it can be
reduced to the essentials with [this] script.

**Warning:** SDE increases the runtime considerably, it should only be used
for small problems.

See merge request [extensions/dune-codegen!328]

  [Intel SDE]: https://software.intel.com/en-us/articles/intel-software-development-emulator/
  [this]: https://bitbucket.org/dwdoerf/stream-ai-example/src/ffccb96ce47c4f3474f1e7f31430c4094479108d/parse-sde.sh?at=master
  [extensions/dune-codegen!328]: gitlab.dune-project.org/extensions/dune-codegen/merge_requests/328