are these files used at all in lifev?

warnings suppression: use the same type for iteration variable and test variable, comment out unused variables

+ removed useless explicit castings

+ commented out unused variables

+ use UInt when it makes sense

In this case, UInt rather than int: after Samuel work on CurrentFE, fe.nbFEDof() and the others "nb*" getters return UInt.

In this case, UInt rather than int: after Samuel work on CurrentFE, fe.nbFEDof() (or fe.nbNode) are UInt.

warning suppression: use the same type for the iteration variable and the test variable (in this case, int rather than UInt)

void source( const UsrFct& fct, ElemVec& elvec, const CurrentFE& fe, Real t, int iblock = 0 ) implementation.

Please, highlight in the commit comment every possible lack of backward compatibility (especially if the commit does not break the compilation, since in this case it is likely to pass unnoticed :-) ). Thanks!

Before the ammissible range was:

abs(defaultOmega)/1024 < abs(Omega) < abs(defaultOmega)*1024

and if Omega was out of the two bounds, it was re-initialized to omegaDefault (which was always positive)

Now we provide (through a new method) two limits: OmegaMin and OmegaMax.
The new ammissible range is:

abs(OmegaMin) < abs(Omega) < abs(OmegaMax)

and if Omega is outside this range, it assume the maximum/minimum ammissible value keeping the sing:
e.g. OmegaMin=0.2, OmegaMax=0.5, OmegaDefault=0.35, Omega=-0.9 ==> New Omega = -0.5 (instead of 0.35).

Fix for the night build: remove the default value for the boost vector that seems to be not "standard" (was working fine on my machine).

The partitionMesh::setup method no longer takes a GetPot data file as a parameter. The number of partitions for the offline partitioning is passed as an unsigned integer:

partitionMesh<Mesh>::setup(UInt partitionNumber, Epetra_Comm &_comm);

Renewed interpolation method:

* Allows interpolation for all the nodal elements
* Code factorization

Changing the definition according to the new constructor.

Adding new methods (usefull for interpolation e.g.)

Adding gradient recovery procedures (ZZ, see documentation)

* Adding the new quadPoint.hpp file in the headers
* Some alphabetical reordering
* Some files were missing (cblas, clapack, dataDarcy, darcySolver), adding them.

Add the new file.

* Adding new constructors / accessors / setters
* Support for all space dimensions (even if it is different from the native space, e.g. 1D quadrature rule in 2D space)
* Adding new features (exactness check, vtk export for debugging,...)
* Documentation
* Backward compatible, no changes in your codes needed.

* Moving the QuadPoint class in separated files (it was in quadRule.hpp/.cpp before).
* Adding new interfaces (fully backward compatible)
* Adding the possibility to use any dimension (not only 3)
* Adding documentation