ci: add GitHub action that automatically creates the derivative databases

[rickstaa]

This PR adds the '.github/workflows/create_db_derivatives.yaml'GitHub action that is used to automatically create the derivative databases (i.e. 'category', 'images' and 'extended') based on the 'plane-alert-db.csv', 'plane_images.txt' and 'blacklist.txt' files.

Files explained

• .github/workflows/create_db_derivatives.yaml: A GitHub action file that runs every time somebody modifies the plane-alert-db.csv, plane_images.txt and blacklist.txt files.
• scripts/create_db_derivatives.py: A file that creates the derivative databases based on the plane-alert-db.csv, plane_images.txt and blacklist.txt files.
• scripts/create_images_reference.py: A small little script that I used to create the new plane_images.txt file. This file can be removed when we are sure that this file is correct.
• plane_images: This is the new planepix.txt file. I kept the file extension as txt to distinguish it from the main database and the derivative databases.
• blacklist.txt: Contains planes that had to be removed because of the new Twitter rules.
• plane-alert-db-extended: The `plane-alert-db.csv' database with the blacklist items added to it.
• plane-alert-db-extended-images: Same as above but now with the image links added.

See the action in action

• ✅ https://github.com/jettracker-org/plane-alert-db/actions/runs/4316391544

How to test

Before switching to this automatic workflow, we have to ensure that the generated databases are equal to the ones already in the repository. I compared them using VScodes compare feature, and the implementation looks correct to me. However, please double-check.

1. Go into the scripts folder.
2. Install the python dependencies.
3. Run the create_images_reference.py script to generate the database of the new image.
4. Run the create_db_derivatives.py script to generate the derivative databases.
5. Compare the generated databases.

Things that could cause problems

My script currently replaces N/a with np.nan. I don't know how this field is read in the https://github.com/kx1t/docker-planefence repository, but otherwise, we have to add those values back in.

Things that were not yet automated

I did not know what was in the following databases:

• badgers-best-images.csv
• badgers-best.csv
• plane-alert-wip.csv
• plane-alert-pia.csv
• plane-alert-search-terms-to-do.csv
• plane-alert-wip.csv

[rickstaa]

Maybe we should change this to only run on pull requests? Now it runs both on the PR and on the main branch.