Improve (and refactor) the operator component

[ema-pe]

After doing a small fix in the operator component (code in the operator/ directory), I noticed some things that should be improved or refactored:

1. There should be some instructions on how to configure and run the operator without forcing the use of Docker to build the image and then run the container.
2. The docker file can be improved: perhaps we should use the official Python docker image, since it already contains some common packages needed to run Python scripts. Alpine is not suitable in this case because the operator leverages the use of Bash and Python, two big dependencies.
3. Python is required to run the plots-results.py script to generate the plots after the tests, but the main execution file is operator_entrypoint.sh: this is a bash script of ~180 lines of code. A large bash script quickly becomes unmaintainable over time, perhaps a rewrite in a real programming language like Python should be done.
4. The user can configure the operator by editing the dfaasoperator.env file and then loading the file that runs the operator with Docker. This forces the user to use Docker (see the first point) and to compress the information in the environment variables, which are usually not suitable for complex and structured information. We should rethink how the configuration can be done: the env file can be kept, but some mechanism is needed to allow specifying the configuration to be used directly (by adding a command line option, for example).