`on_stop` is not automatically triggered with pipelines for merge requests

Closes #60885

See merge request gitlab-org/gitlab-ce!27618

Add basic GraphQL for a Group

Closes #60786

See merge request gitlab-org/gitlab-ce!27492

Due to the nature of pipelines for merge requests, deployments.ref can
be a merge request ref instead of a branch name.

We support the environment auto-stop hook for this case

Fix reviewer roulette when no-one is in a category

See merge request gitlab-org/gitlab-ce!27895

Use correct k8s namespace in Prometheus queries

See merge request gitlab-org/gitlab-ce!27812

Disable HTTP for the nginx-ingress Review App service

See merge request gitlab-org/gitlab-ce!27884

This would return `people.size` before, when it should return `nil`.

Remove pages domain scheduled for removal

See merge request gitlab-org/gitlab-ce!26227

Domain will be removed by verification worker after 1 week
of being disabled

Give reviewer roulette its own header in the docs

See merge request gitlab-org/gitlab-ce!27883

Clarify that reporters can manage boards

Closes #60794

See merge request gitlab-org/gitlab-ce!27616

Docs aren't reviewed using roulette

Closes #61047

See merge request gitlab-org/gitlab-ce!27876

Reviewer roulette via Danger should take OOO status into account

Closes #57652

See merge request gitlab-org/gitlab-ce!27696

Fix letter opener gem error: "undefined method `before_filter' for LetterOpenerWeb::LettersController:Class"

See merge request gitlab-org/gitlab-ce!27769

Refactor LfsImportService and ImportService

See merge request gitlab-org/gitlab-ce!27255

In order to make `LfsImportService` more reusable, we
need to extract the logic inside `ImportService` and
encapsulate it into the service.

Disable method replacement in avatar loading

Closes #60903

See merge request gitlab-org/gitlab-ce!27866

Signed-off-by: Rémy Coutable <remy@rymai.me>

Uninstall cluster applications (backend services)

Closes #60665

See merge request gitlab-org/gitlab-ce!27096

These builder methods are using user provided input inside a public_send
but this is safe to do in this instance because before they are called
we check before calling them that they match an expected application
name.

In order to not miss any errors since we are not checking the return
value of update

Basically does the reverse of after_transition to :installed.

Add endpoint to delete/uninstall a cluster application

Add specs

Only prometheus can be uninstalled atm, the rest will be dealt with
later.

Presumption is that new application types will have uninstallation
implmemented at the same time.

Rescue and put into :uninstall_errored if something goes wrong while
destroying, which can happen. I think it is safe to expose the full
error message from the destroy error.

Remove the :uninstalled state as no longer used.

+ to monitor progress of uninstallation pod