When copying a MatrixBasedOperator, the functors are also copied, in
particular the functor which is appended in the constructors. However,
the copied functor still used the assembled_ member of the
copied-from MatrixBasedOperator, and not the one of the new
MatrixBasedOperator.