Make standalone container registry tests simpler to run

As part of a project horse work item, we were looking to automatically test Container Registry after provisioning. These tests were added in gitlab-org/quality/testcases#1052 (closed) , and it is known that it is a bit more complicated to run the end-to-end tests for Container registry with QA due to needing operational Runner setup.

Problem description

Apart from a known runner requirement, the QA tests for container registry have other limitations. They can either run if:

1. Use an option to run from scratch against an omnibus instance. By invoking https://gitlab.com/gitlab-org/gitlab/-/blob/master/qa/qa/specs/features/browser_ui/5_package/container_registry_omnibus_spec.rb , registry is enabled, runner setup and the tests executed
2. Run against an already setup environment https://gitlab.com/gitlab-org/gitlab/-/blob/443d12193c241fe2de017fdbb81ba377ed8ac552/qa/qa/specs/features/browser_ui/5_package/container_registry_spec.rb#L5

If we select the option of running against a setup environment, the tests only run against staging and pre for gitlab.com environments.

Even when only: { subdomain: %i[staging pre] restriction is removed by editing the code, running the Registry only test is a bit confusing:

bundle exec bin/qa Test::Instance::All https://INSTANCE -- qa/specs/features/browser_ui/5_package/container_registry_spec.rb  --tag reliable 

Proposal

This issue is proposing to make an assumption that if a Runner is setup and enabled on any running instance, we are able to run a QA e2e test on any domain.

Ideally, I'd like us to be able to run all Container registry tests by supplying a tag, eg.:

gitlab-qa Test::Instance::Any VERSION https://example.com --tag orchestrated --tag registry