The call to the nextIteration() method must occur exactly once per
iteration, otherwise incrementing the total iterations counter in
wrong places might cause serious bugs. For example if minIterations is
equal to 1 and the initial residue is smaller than convergenceResidue,
the loop on line 154 in tnlGMRESSolver_impl.h will be entered, but the
m-loop on line 179 will be skipped and on line 263 update() will be
called with empty vectors, which will lead to division by zero.

The condition substituting nextIteration() is different than in the
method itself, because incrementing the iteration counter is avoided.

...and some simplifications in tnlSolverStarter - needs more!

* origin/master:
  Fixing build script.
  Adding OpenMP to vector operations.
  Fixing a bug in host and device arrays comparison.
  Fixing the DICOM reader.
  Fixing DCMTK code to work when DICOM support is not enabled.
  Fixing #1 - tnl-dicom-reader, tnl-image-converter
  Implementing the DICOM reader.
  Implementing the DICOM reader.
  Refactoring code for the DICOM format support.
  Addign DICOM support - TNL can be succesfully build now.
  JPEG format is working.
  Implementing JPEG image write.
  Implementing JPEG image.
  Adding JPEG format support.
  Adding PNG format support.
  Adding conversion from .tnl file to image.
  Implementing image converter.
  Implementing image converter.
  Implementing the image converter.
  Adding image converter.

* master:
  Tuning Ellpack formats.
  Tuning Ellpack format.
  Fixing bug in tnlVector::operator !=.
  Optimizing CUDA L2 norm.
  Adding banchmark for CUDA lp norm.

Conflicts:
	build
	src/core/CMakeLists.txt

* master:
  Optimizing L1 and L2 norms in CUDA.

The system is actually solved for t+tau, not t.

The nextIteration() method should not be called here because it
increments the iteration counter.

* master: (34 commits)
  Fixing compilation with nvcc.
  Fixing C++11 flag in CMakeLists.
  Refactoring tnlFunction.
  Refactoring system for function types.
  Debuging mesh.
  Renaming Ellpack network to indexed multimap.
  Implementing superentity indecis storage in storage network (=indexed multimap)
  Fixing missing superentities of vertices.
  Inserting the Ellpack network to the mesh.
  Renaming Ellpack graph to Ellpack network.
  Adding sparse Ellpack graph.
  Refactoring mesh entity.
  Refactoring the mesh.
  Refactoring the mesh.
  Refactoring the mesh.
  Refactoring the mesh.
  Refactoring the mesh.
  Refactoring the mesh.
  Refactoring the mesh.
  Fixing mesh config validator. Adding mesh simplex.
  ...

This will ensure that in verbose mode, the final state of the linear
system solvers is always printed. Previously this was the case only when
refresh rate was equal to 1.

Conflicts:
	src/mesh/layers/tnlMeshStorageLayer.h
	src/mesh/layers/tnlMeshSubentityStorageLayer.h
	src/mesh/layers/tnlMeshSuperentityStorageLayer.h
	src/mesh/tnlMesh.h
	tests/unit-tests/mesh/tnlMeshTester.h