chore(tech-debt): remoteNamespace expected to exist prior to downstream service deployment

[justinthelaw]

Describe what should be investigated or refactored

The remoteNamespace used by this operator package expects the namespace to exist prior to pg-cluster creation. For example, when a database and secret for litellm needs to be created prior to the deployment of the LiteLLM UDS package.

The workaround seen in most other repositories falls into one of the following practices:

1. Namespace is created by another component whose operator, e.g., MinIO, is capable of standing up a remote namespace prior to its required operations (Gitlab example)
2. Namespace is created using a locally created dev-namespace Zarf package, which is not practical in an airgap and/or requires either a) manual creation of the namespace using uds zarf tools kubectl or b) adding a bunch of extra packages with just a namespace before the postgres-operator is deployed (Mattermost example)

Recommended solutions

Use Zarf onDeploy actions to create namespace(s) as defined in a bundle override for things like secret creation.

OR

Add some documentation in the configuration.md specifying the need for the remoteNamespace to exist prior to postgres-operator creates a remote secret.

Example bundle where we are forced to use dev-namespace-like behavior

https://github.com/defenseunicorns/uds-ai/tree/36-chore-k8s-gpt-bundle-integration