[fix][broker]Namespaces can be created with may empty replication_clusters policy

[poorbarcode]

Motivation

Before #24860, the namespace-level policy: replication_clusters will never be empty; after #24860, it will be empty if a namespace is created with the following request

curl -x PUT http://localhost:50025/admin/v2/namespaces/{tenant}/{namespace}` -D `{"allowed_clusters": ["{current_cluster}"]}`

This will lead to some unexpected problems, such as the namespace not being able to be deleted normally and getting an error Failed to delete namespace {namespace} java.util.concurrent.CompletionException: org.apache.pulsar.broker.web.RestException: Cannot delete the global namespace {namespace}. There are still more than one replication cluster configured.

Modifications

• Recover the default behaviour that was defined before [fix][broker] Fix wrong behaviour when using namespace.allowed_clusters, such as namespace deletion and namespace policies updating #24860
• Improve the error message

Does this pull request potentially affect one of the following parts:

• Dependencies (add or upgrade a dependency)
• The public API
• The schema
• The default values of configurations
• The threading model
• The binary protocol
• The REST endpoints
• The admin CLI options
• The metrics
• Anything that affects deployment

[lhotari]

The test "org.apache.pulsar.broker.admin.AdminApi2Test > testCreateNamespaceWithEmptyReplicationClustersByHttp" fails.