feature: expand e2e coverage across CLI surface

[atkrad]

What problem would this solve?

v1 e2e (#37) covers cluster up, cluster status, and cluster down, plus one deploy/delete smoke per scenario. Commands like plan, logs, resolve, list, and status can still regress against a live cluster without CI catching them.

cluster kubeconfig is also uncovered: v1 reads the kubeconfig path out of the cluster status --output json payload rather than invoking that subcommand, so its output contract is not exercised.

What do you want?

After #37 lands, expand internal/e2e/ (same Kind cluster, same nabattest.Run harness) to cover more of the CLI surface against the live cluster, for example:

• cluster kubeconfig (path form and --raw contents form)
• plan (offline vs live as applicable)
• list / status after deploy
• logs for the smoke workload
• resolve against the fixture

Prefer adding cases on the shared cluster from #37 rather than new Kind create cycles. Keep the golden integration tests for offline rendering.

Note that #37 auto-discovers scenarios from internal/e2e/testdata/, so workload-shaped cases need only a new directory with deployah.yaml plus expect.yaml. Command-surface cases that are not about deployed resources will need their own suite methods.

Depends on: #37

Alternatives considered

• Only unit/integration coverage for these commands - misses live kube/context bugs.
• One giant e2e that hits every flag - too slow and flaky; grow incrementally.

[atkrad]

Note: parent #37 now uses testify/suite (E2ESuite). Expanding CLI coverage means adding methods to that suite (or a sibling suite in the same package). No new cluster create/destroy per command -- they all share SetupSuite's cluster and s.kcPath.