[GDB-14769] Publish the headless service endpoints

[simonzhekoff]

Description

GraphDB's headless Services did not set publishNotReadyAddresses, which defaults to false. This caused two issues:

1. Cluster nodes could not communicate before readiness probes passed, making inter-node discovery impossible during startup.
2. It was not possible to implement rolling-upgrade wait logic in the startup probe, because a not-yet-ready pod would not be reachable by its stable DNS name — blocking proper upgrade sequencing.

Related Issues

GDB-14769

Changes

• Added headlessService.publishNotReadyAddresses: true — allows cluster nodes to resolve each other via DNS before readiness probes pass.
• Added service.publishNotReadyAddresses: false — explicit default for the regular (non-headless) GraphDB Service.
• Added proxy.headlessService.publishNotReadyAddresses: true — same reasoning for the proxy headless Service.
• Added proxy.service.publishNotReadyAddresses: false — explicit default for the regular proxy Service.

Checklist

• I have tested these changes thoroughly
• My code follows the project's coding style
• I have added appropriate comments to my code, especially in complex areas
• All new and existing tests passed locally