feat(core): HTTP readiness probe on admin port 3030

[kkopanidis]

Summary

Core pods were using the same shallow gRPC probe for both liveness and readiness, so Kubernetes could mark Core ready before Redis or required modules were actually serving.

This chart update aligns Core probes with the deep readiness work in Conduit:

• readinessProbe → httpGet /ready on port 3030 (admin HTTP)
• startupProbe → httpGet /ready on port 3030 (enabled by default)
• livenessProbe → unchanged gRPC on port 5000
• Declares admin-http containerPort: 3030
• Sets READY_REQUIRED_MODULES to database,router when router is enabled (database only otherwise)

Depends on: ConduitPlatform/Conduit#1505 — merge Conduit first, then this chart.

Test plan

• helm template renders Core readiness/startup as HTTP /ready:3030 and liveness as gRPC :5000
• Deploy upgraded chart: Core pod stays Running but is NotReady until database + router register
• Kill Redis: Core remains Running, becomes NotReady, recovers without restart when Redis returns
• With router.enabled=false, READY_REQUIRED_MODULES is database only

Follow-up

Router module deep readiness remains out of scope for this change.