Previously, the lexicographical order was used.
This caused 'machine2' > 'machine10'.

Now, when strings contain digits at the same position,
the whole integer are read and compared.
Now, 'machine2' > 'machine10'.

Which means that the first machines are no longer
'machine1', 'machine10', 'machine11' ..., 'machine100' ...
but 'machine1', 'machine2', 'machine3' ...

Old scheduling commands have been replaced with existing scripts.
Batsim Usage is now explicited in the README.
Dependencies versions are now given.
A script to install some dependencies has been added.
Structuration has changed a little.

Previously the energy exported by Batsim was
the total energy used by the machines at the
end of the simulation.

Now the exported energy is the difference
between
  - the total consumed energy by the
computing machines when the last job
is completed, and
  - the total consumed energy by
the computing machines when the first
job has been submitted

find_all_values results were supposed to be
strings but the cast was not doneexplicitely,
which caused the script to fail, e.g. on numbers

The execution scripts now handles properly the
declaration order of variables. To do so, the scripts
find a declaration order and checks that the dependency
graph between variables is acyclic.

Post and pre commands can now be used in the
ExecN instances script.

The test_energy input file of these scripts have been
modified to test those behaviours

Options have been added to:
  - enable warnings
  - treat warnings as errors
  - enable compile-time optimizations
  - disable run-time assertions

gcc/clang versions are checked to ensure they support c++11.
The C++11 flag is set manually to limit CMake version need.
The CMake version needed is now 2.8.11 (target_compile_definitions).

The corresponding output file would be OUT_consumed_energy.csv.
This file is a time series.