Parametrized output folder

[Sinclert]

Problem

One of the issues that is interfering with the next goals that this project repository needs to achieve:

• Be locally runnable by yadage-run.
• Be remotely runnable by REANA.
• Introduce MLFlow parametrization.

is the writing of workflow output files in the same directory as static code and scripts. This causes problems because:

• When locally executing Yadage, there can only be one workdir. The "workdir" is basically the folder where workflow steps can interchange files with each other. Now, due to this project's code and script into two independent folders (docker-madminer-ml and docker-madminer-physics), there were actually two "workdirs", which was invalid.
• When executing on REANA, it is very common to differentiate folders due to read/write permissions. Having output files being written were the code and scripts are stored blurs the barrier between only readable and readable & writable folders.

Proposed solution

Thanks to the definition of an intermediate shell-script layer between the workflow specification and the Python code, the output folder is now parametrizable.

This PR introduces a new argument (-o) to all the scripts (both in docker-madminer-ml and docker-madminer-physics folders), which is used as the path to store files.

Other additions:

• Re-define scripts permissions to 755 (otherwise non-executable).
• Remove __future__ syntax from Python3 files, which is Python2 legacy.
• Define a root folder Makefile, which simplifies the local execution of yadage-run by:
  • Specifying the correct initdir option (for the inputs.yml file).
  • Specifying the correct rootdir option (for the workflow.yml and steps.yml files).

The definition of initdir and rootdir was only possible after a private conversation with @lukasheinrich, as they were not explained in the Official Yadage documentation (will they soon be? 🤷‍♂️ )

[irinaespejo]

Thank you @Sinclert , this is the PR we talked about over a private Zoom meeting. I tested it locally and works smoothly until the Pythia step. I think we can proceed with the merge. Thank you this is a crucial PR!