Zubbi

Fixes
- How-to page is updated with description of supported_os and reusable directives.

New Features
- **UI:** Improve search results ordering. The search results will be ordered by
the priority of the fields matching the query. The priority is like the following:
name -> description -> tenant/repo.
- **Scraper** Projects(repositories) can be configured as "reusable" in settings.cfg.
When A repository is configured as "reusable", all jobs and roles scraped from this
repository are marked as "reusable".

New Features
- **Scraper:** Support marking roles and jobs as "reusable". It parses the directive
in role's README file and job's description and store it in Elasticsearch.
- **UI:** Search result will display roles and jobs that are marked as "reusable"
on top, and highlight them.

General
- Updated dependencies to newest versions

Fixes
- Fixed a bug where the delete outdated query never matched repositories on
Github. To fix this behaviour, an additional repo_name.keyword field was
introduced in the git-repos index.
- The connections are now first initialized when a valid scraper command is
invoked. Thus, running `zubbi-scraper --help` or providing wrong command line
arguments should directly run and return without unnecessarily initializing
the connections to Github, Gerrit and Elasticsearch.

General
- Updated dependencies to newest versions