chore: test refactoring to use separate environments per spec

[msuperina]

This is a first attempt to have tests in a bdd style with an environment injected at will (like in the beforeAll / beforeEach hooks). Some commands are tagged as "setup". At the moment it is assumed that there is only one "not setup" command per test, but this can change in the future. This will need refactoring, as a setup command will become a "setup" for the closest previous "not setup" command.