Zubbi

Fixes
- Fixed a bug that made zubbi-scraper crash on empty repositories.

Fixes
- Filter job files for *.yaml extension before parsing to avoid yaml parser
errors. So far, Zubbi was parsing all files (including e.g. a README.md file)
within the jobs directory which made the parser fail.
- Make the RepoParser more robust against files with invalid yaml content. In
case an error occurs during job or role parsing this should now only affect
the current repository but not the whole scraper run.

Fixes
- Remove the noreferrer header from the zubbi repository link in the footer
to not leak internal URLs.
- Use pagination when fetching installation ids from GitHub. So far, Zubbi
only fetched the first 50 installations.
- Ignore job file if there is syntax error when parsing yaml content.

Fixes
- Fixed a bug that resulted in an internal server error when using a search
query that contains a '/' in zubbi-web.

Fixes
- Fixed a bug that made the zubbi-scraper crash when a zuul config directory
contained not only files but a nested directory structure. Those config
directories are now scraped recursively.
- Fixed a bug that made zubbi-scraper crash on startup if the TENANT_SOURCES_REPO
setting referenced an unknown connection.

Fixes
- Fixed a GraphQL query which was relying on a master branch being present in
the repository.