test: add e2e tests

[AlexanderLanin]

This adds the first E2E tests for otterdog.

Most of the tool’s logic only really shows its true colors when different parts interact, so starting with E2E tests feels more useful than writing isolated unit tests. These tests check if the whole system behaves correctly, not just if individual pieces pass in isolation.

The main goal is that these tests stay stable even if the internal code changes. As long as the user-facing behavior is unchanged, the tests should continue to pass.

These tests run the tool as a pure black box, only through the public interface. The trade-offs:

• every new feature needs to be reflected in the GitHub mock
• error localization is harder, since failures happen at the outer boundary
  → in practice, the existing -vvv flag helps a lot here

Happy to hear what you think about the direction!

[netomi]

I also favor e2e or integration tests over unit tests. Do to the nature of the tool, you have to mock a lot of stuff which I tried to avoid as much as possible so far. Doing things similar to PyGitHub might make sense, where they record responses from GitHub and replay them during tests.

Will take a look at the PR and play aound with it.

[AlexanderLanin]

No overlap with #567 But this needs to either:

• use the API-expect logic from test(integration): add integration tests for model-to-GitHub apply flow #567, which would make it overlap a lot
• or it needs to stop at the model level and verify that patch creation was started, to not overlap with test(integration): add integration tests for model-to-GitHub apply flow #567 at all

[AlexanderLanin]

Not working on this any further at the moment, so let me close this.