CLI needs integration/e2e testing

We should add a suite of tests that test the different use cases of the CLI.

Currently this is only done implicitly via things like paramedic (which is used by all plugin CI) or mobile-spec (which is run manually on releases), which use the CLI to do things. But I doubt this really covers all the cases and probably mostly also only uses already release CLI versions.

Start by reviewing the CLI use cases covered indirectly through paramedic and mobile-spec, including how those tools invoke released CLI versions. Define the missing integration or end-to-end scenarios and the test entry points before implementing a suite whose coverage demonstrates the CLI works across its intended use cases.

Written by the indexing model from the issue text.