refactor: simplify extractor

[cmdoret]

This PR simplifies the extractor interface by adding a unique Repository model (see #71).
Until now, all repository metadata was stored directly in extractors, and each extractor had to define its own schema for RDF serialization. The aim of this PR is to decouple the extractor from the repository. It does the following:

• Extractors only have the (public) methods list_files() and extract().
• extract() returns a Repository object (instead of editing the extractor in place.
• Repository is a dataclass that stores the repository metadata and can be serialized to RDF.
• All serialization schemas are removed, as we only need RepositorySchema

Using a single serialization for all extractors required minor metadata alignments. I made the following choices:

• There can be multiple authors (git and github had a single author, whereas gitlab had multiple)
• Dropped schema:isPartOf, which was only used in GitLab
  • We also used it to denote affiliation to an organization, which is not its intended use

[rmfranken]

tres beau.

[rmfranken]

So safe!