Add code coverage measurement for the Python and Bash test suites

[codeforester]

Neither requirements-dev.txt, pytest.ini, nor any CI workflow references coverage/pytest-cov, and there is no equivalent for the Bash/BATS side. With 930+ Python tests and 50 BATS files, "what fraction of repo.sh or base_projects/engine.py is actually exercised" is currently unanswerable without a manual audit.

Scope:

• Add pytest-cov to requirements-dev.txt and wire coverage reporting into the python CI job.
• Establish an initial coverage floor (even a loose one) so regressions are visible; ratchet over time.
• Investigate a lightweight equivalent for Bash (e.g. kcov or bashcov) as a follow-up if a Python-only floor proves insufficient signal for the Bash-heavy command layer.

Acceptance criteria:

• CI reports Python coverage and fails below a documented floor.
• The floor and rationale are documented in docs/testing.md.