Rework recursive process_file core calls

[kukovecz]

Currently we have the unblob running around recursively calling the process_file function.

• This makes the investigations and extending the core code harder.
• In this way the parameters of the process_file are mixed: it contains variables, configuration globals, local states, etc....

We had an idea that instead of recursively calling it over and over again, we could have a simple Queue, where we could put Tasks which needs to be processed. This would potentially improve the code quality and add the ability to easily extend with multiprocessing.

Somehow related to:

• Parallelize file processing #71
• Set up performance testing #100
• Make unblob usable as a library #101