Add pre-commit hooks to this repo

[anakin87]

This PR is for adding pre-commit hooks to the tutorials repository.

• I copied .pre-commit-config.yaml from the haystack repo, then I removed all the hooks not related to Jupyter notebooks
  (I understood that pure python tutorials won't be supported)
• Tested on Ubuntu 22.04

Some notes/questions:

• (If I try to run pre-commit run --all-files, it seems that black-jupyter (22.6.0) wants to reformat all the notebooks. It can be an issue of my environment.)
• If we want to add more checks for Jupyter notebooks in the future, we can use some of these
• The contributors' guide should be updated. Better to open another PR?

[masci]

Let's change the contributors guidelines in this PR.

@TuanaCelik we never run a formatter on the notebooks, I think this is a good opportunity to do it. The only problem I see, it's already hard to port changes from the haystack repo now, if we re-format the files it'll be impossible. So I would re-format all the things only if we're close enough to remove the files from the haystack repo. WDYT?

[TuanaCelik]

@masci - regarding formatting, sounds good to me. Let me make sure I've got it right:

• We would reformat the tutorials but not yet so that it's still easy to pull current changes to tutorials from haystack
• Once we're about to pull the plug and move to this haystack-tutorials we would reformat and stop any further changes in tutorials in haystack

Correct?

[anakin87]

• Put back check-ast hook for python syntax, since there are few python scripts in this repo.

• Moved requirements.txt to root and added pre-commit to it.

• Added a very minimal section in Contributing.md about environment preparation and pre-commit.
  We could better explain all the preparation of the environment as it happens in the main repository; however, reading the current guide, my impression is that in this repo we want to favor simplicity... @TuanaCelik @masci WDYT?

[masci]

Thanks for the PR! Can we also invoke python /scripts/generate_markdowns.py in the pre-commit hook? The generation is pretty fast, we can afford to run it at every commit I think.

[anakin87]

Thanks for the PR! Can we also invoke python /scripts/generate_markdowns.py in the pre-commit hook? The generation is pretty fast, we can afford to run it at every commit I think.

I agree.
I had thought of doing it in another PR, but then I continue on this one and resume the work I had done in deepset-ai/haystack#3192.

[anakin87]

Taking inspiration from deepset-ai/haystack#2979, I refactored generate_markdowns.py:
it is invoked by a custom pre-commit hook and if a notebook tutorial is created/changed,
it generates the corresponding markdown file.

It is still possible to run the script on all the notebooks, using the argument --all.

I tested it on Ubuntu 22.04.

[masci]

Hook doesn't work on mac, otherwise looks good

[masci]

This doesn't seem to work on macOS, filepath is relative to the root of the repo while notebook_tutorials_dir is absolute.

I had the same problem in Haystack, I think we can hardcode paths here and use relatives everywhere.