Compatibility with recent version of Python, Tensorflow, etc.

Objectives:

• Create a table with an overview of the supported versions of Python and Tensorflow. Do want to include any other dependencies? Perhaps Numpy?
• Include above table in the official Ketos documentation
• Ensure Ketos actually supports these versions
• Create multiple Docker images to test support across versions
  • For example, we could create a separate Dockerfiles for each supported version of Python with the latest compatible version of Tensorflow, as detailed here: https://www.tensorflow.org/install/source

Python	Tensorflow
3.8	2.13
3.9	2.17
3.10	2.17
3.11	2.17
3.12	2.17

• Modify CI pipeline to test against all the Docker images
  • I'm not sure how best to do this, but a straightforward solution would be to simply repeat the test section of the CI script multiple times, each time pulling a different image, as discussed here: https://forum.gitlab.com/t/using-different-docker-images-during-a-single-gitlab-ci-job/22938
  • To not use excessive CI time, we would only test against all images on the main/master branch, whereas all other branches would only run tests against the latest image

@frazao @padovese , your thoughts?