- the spaceData enum starts from 1 (rather than the default 0), to avoid crash
  if the string passed to setSpace does not match one of the stored ones
- spaceData and M_spaceMap now include also P0 elements and P2Bubble elements
  (the latter valid only in 3D)
- private becomes protected. This is needed in the ecm2 module's FESpaceManifold
  which we plan to bring soon into lifev
- some return values in the getters are turned into references rather than values
- changed a comment to a method, which referred to a vector as a velocity vector
  (problem specific)

Bug fix in StringUtility.cpp: we were potentially writing/reading out of a char array. Also, the number of digits of an integer i is not i/10, but Log(i).

Modified the interface of the PartitionIO and HDF5IO classes
to use Epetra_MpiComm objects explicitly. These classes are
only enabled if LifeV is configured with MPI and HDF5 and they
are only used with MPI.

If the user of the classes need to create an Epetra_Comm object
in his code, he should explicitly cast it to an Epetra_MpiComm
when using the PartitionIO class.

Replacing the test on the value of a pointer after a dynamic
cast with an assert. The dynamic cast should not fail as the
pointer should always always be an Epetra_MpiComm object.

 removed the warning "comparison between signed and unsigned integer expressions" in DOFInterface3Dto3D

The files HDF5IO.*pp and PartitionIO.hpp are not included
in compilation when TPL_ENABLE_HDF5 is off.

However, to prevent unforeseen problems with external apps using
LifeV, the LIFEV_HAS_HDF5 include guard has been added to these
files.

This commit registers HDF5 as an optional direct dependency of
LifeV. If enabled, the preprocessors symbol LIFEV_HAS_HDF5 is
defined in Core_config.h

The files HDF5IO.hpp, HDF5IO.cpp and PartitionIO.hpp, as well
as the partition_io test in the Core testsuite are only enabled
and compiled if the HDF5 TPL is enabled.

The HDF5 functionality in ExporterHDF5, which uses the
EpetraExt::HDF5 is (for now) unaffected by the activation or
deactivation of HDF5 as a LifeV TPL, since it depends on
HDF5 as a Trilinos TPL.

The merge of the astyle commits was done improperly
and PartitionIO.hpp had syntax errors

structure_flag is now solid_flag

for constructing the rotation matrix (and its transpose) we don't need the
whole system matrix, we just need its row map.

Conflicts:
	lifev/core/fem/BCManage.hpp
	lifev/core/fem/BCManageNormal.hpp

nbLocalCoor, for clarity purposes. The getter nbCoor() is left as
deprecated for backward compatibility reasons (should be deleted in the
next major release).

This is true for the nbCoor method/attribute in both CurrentFE (and its
derived) and ReferenceElement (and its derived) classes.

Also, in CurrentFE.cpp the barycenter method has been slightly changed:
now it cycles over ALL the components of the cell nodes (like it was
already done for the diameter() and diameter2() methods ), so that in case
the element is 2D but on a general plane, the barycenter is computed
correctly. Otherwise we would need to override the method inside
CurrentFEManifold.

Conflicts:
    lifev/core/fem/AssemblyElemental.cpp
    lifev/core/fem/AssemblyElemental.hpp
    lifev/core/fem/CurrentFE.cpp
    lifev/core/fem/CurrentFE.hpp
    lifev/core/fem/CurrentFEManifold.cpp
    lifev/core/fem/CurrentFEManifold.hpp
    lifev/core/fem/DOFLocalPattern.cpp
    lifev/core/fem/DOFLocalPattern.hpp
    lifev/core/fem/GeometricMap.cpp
    lifev/core/fem/GeometricMap.hpp
    lifev/core/fem/ReferenceElement.cpp
    lifev/core/fem/ReferenceElement.hpp
    lifev/core/fem/ReferenceFE.cpp
    lifev/core/fem/ReferenceFE.hpp
    lifev/core/fem/ReferenceFEHdiv.cpp
    lifev/core/fem/ReferenceFEHybrid.cpp
    lifev/core/fem/ReferenceFEHybrid.hpp
    lifev/core/fem/ReferenceFEScalar.cpp
    lifev/core/fem/ReferenceFEScalar.hpp
    lifev/core/fem/SobolevNorms.hpp
    lifev/heart/solver/HeartStiffnessFibers.hpp
    lifev/navier_stokes/solver/StabilizationIP.hpp
    lifev/navier_stokes/solver/StabilizationSD.hpp

  - Name changed from CurrentBoundaryFE to CurrentFEManifold
  - Reordered files in core/fem/CMakeList.txt so to be in alphabetical
    order (we should do this in every CMakeList)
  - Moving a couple of include out of BCBase (they're not needed there)
  - Adding the transpose of the rotation matrix in BCManageNormal. This
    was already done a while a go, but it got lost somehow. Also, using
    the transpose in the coordinates shift to improve performance, and
    added a getter for the two rotation matrices (I think it's necessary to
    manually zero-out rows of gradient matrix in NS equations in presence
    of directional BC's)
  - Purgeing a couple of functions that were declared but not defined, one
    in BCManage.hpp and one in BCManageNormal.hpp
  - Changeing a bit the interface of the build method in BCManageNormal,
    passing a MapEpetra rather than the whole matrix, and not passing the comm
    object (which can be retrieved from the map).

Conflicts:
    lifev/core/fem/AssemblyElemental.cpp
    lifev/core/fem/AssemblyElemental.hpp
    lifev/core/fem/BCHandler.hpp
    lifev/core/fem/BCManage.hpp
    lifev/core/fem/BCManageNormal.hpp
    lifev/core/fem/CMakeLists.txt
    lifev/core/fem/CurrentFEManifold.cpp
    lifev/core/fem/CurrentFEManifold.hpp
    lifev/core/fem/DOFInterface3Dto3D.hpp
    lifev/core/fem/FEDefinitions.cpp
    lifev/core/fem/FESpace.hpp
    lifev/core/fem/ReferenceFEHybrid.cpp
    lifev/core/fem/ReferenceFEHybrid.hpp
    lifev/core/mesh/MeshChecks.hpp
    lifev/core/solver/ADRAssemblerIP.hpp
    lifev/navier_stokes/solver/StabilizationIP.hpp
    lifev/structure/solver/VenantKirchhoffElasticHandler.hpp

Conflicts:
	lifev/fsi/testsuite/fsi_restart/fluidGI.h5
	lifev/fsi/testsuite/fsi_restart/solidGI.h5
	lifev/multiscale/testsuite/multiscale/CMakeLists.txt