Generate annotated Python code as output

[cristianrcv]

The main contribution of this PR is to generate Python code as output (a part from C and Fortran). The generated Python code is annotated in a OMP fashion (using comments of the form # parallel for PRIVATE() REDUCTION()) and can be executed sequentially. However, since Python lacks of a standard parallel annotation the comments are not processed to execute it in parallel (they are just comments). The main modifications to achieve this involve adding a new constant for Python language, modifying the pretty printer. Notice also that the INDENT_STEP has been changed to 4 because it has no effect on C/Fortran code but it does in Python (it is the default indent step for Python).

The rest of the contributions refer to:

• Renaming the get_submodules.sh script to submodules_get.sh for naming convention with respect to the other submodule script and other PERISCOP repositories. Make script fail on any error.
• Update the get_submodule.sh call in README to submodule_get.sh.
• Make the autogen.sh script fail on error, add verbose on autoreconf command, and do not create the autoconf folder since it is already created by the autoreconf tool. This modifications are introduced to easily detect build errors when building CLooG in a larger pipeline.
• Make the genversion.sh script fail on any error.
• Adding some build files to the .gitignore.
• Ignore untracked files on submodules (mainly because of generated files after building)

[cristianrcv]

Rebased fork to fix merge conflicts.

[cristianrcv]

As discussed in other PRs for PERISCOP tools (https://github.com/periscop/candl/pull/14), I have removed the submodules_update.sh script.

[cristianrcv]

@harenome Could you please consider reviewing this PR since it is the only one that I have submitted that really matters to me?

[cristianrcv]

@Ced Could you please consider reviewing this PR?

[cristianrcv]

I use PLUTO with ISL version 0.18 so I cannot use the last CLooG version. Reverting to 0.18.5 and closing MR.