(re-)Add Coverage support
At the moment, Coverage is not formally included in the development dependencies and is not run regularly to check the coverage of the test suite. An update to Avocado at some point regressed the Makefile script, and I've yet to return to the subproject.
From memory, an issue is that the new avocado runner creates an intermediary python process through which coverage.py is not suitably initialized, so the main Coverage runner script reports 0% coverage.
Open questions:
- Should there be a pass/fail on the total coverage percent? Should a GitLab job fail if it falls below some threshold?
- Are there any integrations available that would integrate well with GitLab better than a simple pass/fail indicator? Badges, Widgets that appear on MR summary pages, robust reporting of html coverage reports, etc?