I started running the test suite in a different way and found that some tests failed because "couldn't run `apt`". That was totally a leak, and tests running apt scare me. So I fixed/simplified them. In detail: - `test_parts.TestPartsLifecycle.test_run_actions_progress`: used a "null plugin"; using the charm one was irrelevant for what is being tested (that progress was informed for the different actions). - `test_build.test_build_basic_complete_structure`: this was more like an "integration test" from the very first days, it doesn't make much sense since the real charm building is responsibility of the `charm_builder.py` script, and it was patched more and more in the last months, and this was the final nail in the coffin. Its conceptual flaw is that it uses the external machinery and then check that the charm built ok, which of course needs *a real underlying execution*, which of course is not 100% possible in a test. - `test_build.test_build_using_linters_attributes`: improved the test to use a better project setup (copied from other tests), which even lead to a minor simplification.