add GCC pragma for trilinos inclusions

use numeric_limits<Real>::quiet_Nan( ) instead of "0/0".

Conflicts:
	life/lifefilters/ExporterHDF5.hpp

Always specify the namespace std when calling abs or fabs

const flag_Type UPDATE_PHI(UPDATE_ONLY_PHI);

to

const flag_Type UPDATE_PHI(UPDATE_ONLY_PHI
                           |UPDATE_ONLY_CELL_NODES);

This commit aims to fix a subtle bug in the call to the function abs().

According to the include header (cmath or cstdlib) the function
int abs(int) has an overloaded function double abs(double) (in cmath) or not.

As a result, if the wrong header is included (i.e. if cstdlib is included instead of cmath)
then a down cast from double to int may happen when abs(x) is called,
introducing a bug.

A clean solution, implemeted in this commit is to always use:
int std::abs(int) when working with integers
double std::fabs(double) when working with doubles.

it tells the importer how many different files to look for when
reading a single variable (it's usually the number of processes
that produced the files in the first place)

let the exporter be a temporary variable (not a private member)

each process will check whether it owns the globalId of the DOF read from
file. If so, it will modify the ExporterData in the corresponding entry.
Other cosmetic fixes.

+ simplified the interface of the read*Data methods (now only receiving
a line with the data to be extracted, and the vector where to put the
values - plus an extra param in the binary case)

If the mesh is the original one, the member _pointList is empty,
therefore we should call the point(i) method.

This is due to compatibility reasons with the exporters, to have the
same interface as RegionMesh3D.

The restart of a simulation should be done saving the initial
mesh, without any deformation.

could become a template for a generic TestExporter

1. the constructor is not empty (will write a VTK collection in a pvd file)
2. rename createMaps -> createPointsMaps (this is what the method really does)
3. new method composeVTKCollection
4. composeVTUGeoStream does not require localToGlobalPointsMap any more
5. first steps towards the Import functionality (readVTUFiles, readBinaryData)
6. minor adjustments here and there

this method is really acting on the current time index rather than on the start
index

Conflicts:
	life/lifesolver/OseenSolver.hpp

precision.

Adding SLV- flag to output lines: this help finding SoLVers lines in
the log.

These quantites should not be read from the data file at this
time otherwise they may overwrite the modifications performed in the
data container, which has been already filled.

Until Paolo Tricerri finishes his development on the structural solver
I commit a temporary fix to restore the full functionality for the
case of 1 single material. In particular:
 - A new boolean variable M_materialsFlagSet has been introduced. It
is set to false if the string "material_flag" is not provided in the
data file, while it is true in the other case.
 - If a material flag is provided, the data container works as before.
 - If no material flag is provided, the data container store the Young
modulus and the Poisson in ID = 1. The getters use the
M_materialFlagSet variable to understand if they have to provide the
quantities for the user provided material or if they should return the
values stored in ID = 1.
 
 This approach is compatible with old code (no changes have been done
in the testsuite), but it allows more flexibility (especially for the
Multiscale framework).

Additional minor changes to the class:
* The variables M_factor, M_verbose, M_order, and M_solidType were
wrongly setted due to missing "section" in the getpot string;
* Inlined some setters and getters;
* Some cleaning and renaming;