[DOC] Relevant in-depth tutorial on FCI

[adam2392]

Brought up in #106 (comment), we want to develop a good example (or set of examples) of causal graphs that we then generate data from that we then feed into FCI.

This can be then used to illustrate how FCI applies its rules and that selection bias (i.e. undirected edges), bidirected edges and directed edges can all be recovered in certain settings. We can both use an oracle and real CI test to further illustrate the impact of faithfulness assumption.

Note: compared to dowhy, we want to run all examples in every single CI run, so ideally the full example should be runnable on your own computer/laptop within <20 seconds. If this is not feasible due to the running of the CI tests, we can add hacks, such as:

• create a code-block section that is not run (e.g. could be just commented out, or in a "Note" block), but if run, generates the corresponding PAG. We can save the PAG to disc and keep it in the same folder as the relevant tutorial.