Correct getting-started instructions and verify deployed database behavior #2051

What content needs to be created or modified?

The overview and deployment page still promise Redis although the connection tutorial uses PostgreSQL and an authored Secret. Commands assume a namespace and do not wait for both current rollouts.

Describe the solution you'd like

Replace stale Redis text with PostgreSQL/Secret descriptions. Use kubectl config use-context when selecting a context. Find the actual runtime namespaces using application/resource labels rather than assuming the Environment namespace. Scope kubectl commands to those namespaces and wait for both demo and database deployments.

Add an end-to-end check: create and complete a todo, refresh, verify its stored SQL row, delete it, and verify the row is absent. If documenting transport verification, inspect the current demo pod's pg_stat_ssl session rather than a separate client; an empty result must not count as success. The demo closes its pool after each request, so provide a reliable way to observe a live session.

Verify commands against the selected supported CLI. A local test with CLI commit 44eaf65 found that initialize, application graph, and application delete reject the tutorial's --preview flag. Do not replace flags globally without checking the supported version.

Where should the new material be placed?

docs/content/getting-started/_index.md and the install-radius, deploy-demo and add-connection pages.

The associated pull request from the Radius code repo, if applicable

N/A. Split from #2041.

Additional context

The previous local candidate run verified namespace discovery, both rollouts, HTTP CRUD, persisted rows and the actual demo connection. It required supported-command adaptations and does not prove the unchanged public tutorial works.