Questions RE: contributing to testing of Manim

Hi. [In the `README` it mentions that](https://github.com/3b1b/manim/blob/b163b1b685445dd38dd0e8e300ad567e037c6f5a/README.md#contributing)

> there is a dire need for tests

As I'm currently moving a quite a few of my own projects CI systems to GitHub Actions I was wondering if there is any general consensus or thoughts on how testing on Manim should go. Obviously animation is a tricky thing to test rigorously, but at least installation and some other basic tests can be started on.

I'm generally a fan of [pytest](https://docs.pytest.org/en/latest/), so unless there are strong objections if I were to start a testing PR on this before the end of the year would people be fine using pytest? Regardless, I have a question related to how to treat Manim in testing. Is Manim meant to be a library or an application? This may sound vauge, so I will just let Donald Stufft's post "[`setup.py` vs `requirements.txt`](https://caremad.io/posts/2013/07/setup-vs-requirement/)" and the [`pipenv` website](https://pipenv.readthedocs.io/en/latest/advanced/#pipfile-vs-setup-py) explain what I'm asking and why for me better than I can.

Thanks in general for all this nice work!

cc @3b1b @eulertour