How well do Jupytext and `plainb` agree on the Jupytext test files?

Following the presentation of jupyter-plainb by @Yahiewi at https://notebook.link/@Yahiewi/demyst , I think we need to ensure that the two parsers understand text notebooks in a comparable way.

The objective of this issue are the following:

1. evaluate the current agreement
2. see how we can ensure consistency over time.