fix: allow all egress from RHDH pods in CI

[rm3l]

Summary

The RHDH chart 2.y ships default-deny NetworkPolicies that only allow egress on specific ports (443, 53/5353, 5432, 6379). E2E tests talk to services on many other ports — Keycloak HTTP routes, ArgoCD, external APIs — causing failures like RPError: outgoing request timed out when the chart's NPs block the traffic.

This PR allows all egress from RHDH pods in CI by applying a permissive NetworkPolicy after deployment. A TODO is kept to narrow this down to explicit ports once all required destinations are mapped.

Changes

• Added NetworkingV1Api to KubernetesClientHelper with an applyNetworkPolicy method (same create-or-replace pattern as ConfigMap/Secret)
• Applied a rhdh-allow-all-egress NetworkPolicy after RHDH deployment
• Used the correct pod selector labels per deployment method (Helm: app.kubernetes.io/name + instance + component; Operator: rhdh.redhat.com/app)

Why allow-all instead of specific ports

Keycloak in CI uses a plain HTTP Route (port 80) accessed via the external Route URL, which resolves to a router/LB IP — not a pod IP. The chart's in-cluster port 80 NP uses namespaceSelector: {}, which only matches pod IPs in namespaces, so it doesn't cover Route-based access. Beyond Keycloak, other E2E services also use non-standard ports. Rather than chasing individual port gaps, we allow all egress in CI and plan to narrow it down later.

Test plan

• Verify workspaces using Keycloak auth no longer fail with OIDC timeout
• Verify the NetworkPolicy uses correct pod selector labels for the deployment method
• Baseline with allow-all-egress confirmed ~40 fewer failures vs port-80-only