* use simpler setMatrix() method for SolverPETSC
* prepared everything to make test fail also in the case of a solver
  that does not converge

  because the implementation does not exist!
* adapted comments about use of CSRmat2MSRmat to reality and made available
  in API documentation

of a nonlinear system of equation.
The aim is to use methods deriving from domain decomposition and apply them to
a soupled non-linear system.

the class generalizedAitken is used by noLinRichardson to determine the
relaxation parameter. It is wrote in general form, even if it still lack of
some codelines for the real general case. Should work for nonLinRichardson.

   LifeV::StaticBdFE::invArea'

finalizing PETSc automatically. Should be reworked more cleanly...

* renamed PETSCforSingleton to PETSC and moved GetPot functionality to above
  function
* make use of singleton pattern

* replaced implementation of CSRMatr::diagonalize_row(...)
* improved documentation

the mass_Hdiv are modified: one should provide the INVERSE of the permeability to all the mass_Hdiv methods.
a new one (based on a functionnal permeability) is added.

operator TP_VdotN_Hdiv improved. works for tetra now.

modifications to allow the creation of RT0 mixed hybrid finite elements
for the TETRAHEDRA.
work performed by a student of Alessandro's that I incorpored.

ensight7-write.hpp to lifefilters and renamed it
ensight7Writer.hpp. fixed compilation for MP solvers ;)

around
sanitize the file name :)
added proper license and authorship

Fernandez -> JFG

update \todo list

dataTime was already used in some solvers and there were some
incompatibilities with what I wanted to do. For example, _inittime and _endtime
are in dataNavierStokes whereas I think they should be in dataTime.
I guess we should merge dataTime and dataTransient and remove the "time"
data from the Navier-Stokes data.

we make a symlink in the local directory
now test_mass_coupling is operational

and change solverPetsc accordingly to select the proper branch