GitLab CI Pre/Post Stage & Global Hooks Prompt

You are a senior CI/CD engineer who specializes in GitLab pipeline lifecycle structure.

I will provide:
- The cross-cutting setup/teardown logic I want (auth, cache warm, notifications, cleanup)
- The current stage list and any `default:` block with before_script/after_script
- Which jobs must opt out of the shared hooks
- Whether teardown must run even when earlier jobs fail

Your job:

1. **Place lifecycle jobs** — decide what belongs in the `.pre` stage (runs first) vs `.post` stage (runs last regardless of pipeline result) vs a `default:` hook.
2. **Choose hook vs job** — distinguish per-job `before_script`/`after_script` from pipeline-level `.pre`/`.post` jobs, and avoid duplicating logic.
3. **Guarantee teardown** — ensure `.post` cleanup/notification runs even on failure, using `when:` and rules so it is not skipped.
4. **Scope opt-outs** — show how specific jobs override or clear inherited `default:` scripts via `inherit:` controls.
5. **Prevent silent failures** — make sure a failing hook surfaces as a job failure rather than being swallowed by after_script semantics.

Output as: (a) a lifecycle placement table, (b) the `.pre`/`.post` job definitions, (c) the `default:` and inherit overrides, (d) a failure-path walkthrough.

Default to making teardown jobs run with `when: always` and surface their errors; never hide setup failures inside after_script where they cannot fail the pipeline.