Fewer dependencies for the container image

[oyvindhagberg]

This PR does the following:

• The entrypoint in the container image will create the CITEXT extension in the database if needed. Previously, that had to be done before starting the container.
• The container image is built with default mregsite and logs directories, so it can run without these being mounted into it. Previously it would exit unless these were mounted in.
• A docker-compose.yml file is added to the repo.
• The section about Docker in the README file is updated. Also add names of more contributors.
• ~The container-image.yml workflow is modified to trigger on pushes to all branches, but only push the built image to ghcr.io if the branch is master. This way, we get to test that workflow on new code before merging into master.~ Done in https://github.com/unioslo/mreg/commit/749d0968b034ee982c91ee17d37df3d04e3262c7

[coveralls]

Coverage: 98.864%. Remained the same when pulling 121566ee99dc1804942d87f136fe9dab8307624a on self-reliant-container-2023w13 into 749d0968b034ee982c91ee17d37df3d04e3262c7 on master.

[terjekv]

This got me thinking... Should we name and/or tag the artifacts better? Naming them something like release and master, while tagging with release number and commit hash, so it's a bit clearer what container one is running?

[oyvindhagberg]

This got me thinking... Should we name and/or tag the artifacts better? Naming them something like release and master, while tagging with release number and commit hash, so it's a bit clearer what container one is running?

Good idea, let's do that, but not in this pull request. I created an issue