This class was not migrated before, not it should be all right

In particular removed  solveSystem which i now in SolverTrilinos.
Moreover the data file should be modified accourding to:

reuse prec and max iter for reuse have to be set in the datafile:
    M_maxIterForReuse = dfile( ( section + "/max_iter_reuse").data(), M_maxIter*8/10);
    M_reusePreconditioner = dfile( (section + "/reuse").data(), M_reusePreconditioner);

compilation warnings suppression

check for EPETRA_MPI symbol after #inclusion of Epetra_ConfigDefs.h

Fix some bugs in the assembly. Enable IP stabilization.

[time_discretization] - for the time

[space_discretization] - for the space

Also other minor changes..

Modified AdvectionDiffusionReactionSolver in order to handle the 2D case. The ip stabilization is compiling and running but should be checked, (also in the 3D case) since preliminary tests revealed some inconsistencies. The sd stabilization wasn't working with different FEspace for the u and beta; it has been fixed but should be double checked. The other Solvers are not compiled when TWODIM is defined.

I think I have to modify also the main.cpp of each test to include the factory.

notes:
* from now on you will need trilinos instead of aztecoo
* the testsuite is now broken.
* the only solvers that works for now are Oseen, VenantKirchoff and fsi.
* the Handler classes are now obsolete. They are superseeded by FESpace
* the postprocess should come out of the solvers and the default filter
  is ensight. I suggest to install paraview 3.x

next steps:
* import trilinos in cmcsforge
* clean-up the testsuite
* insert mpirun in tests executions
* install trilinos on cmcsforge

Let lifev-dev know of any problem, please