fix(pmm): set gRPC service to ClusterIP to avoid duplicate nodePort conflict

[somaz94]

The -grpc service created when ingress.enabled=true inherited the main
service.type. With service.type: NodePort and explicit
service.ports[].nodePort values, both monitoring-service and
monitoring-service-grpc requested the same nodePorts, so the second service
failed to deploy (a nodePort can only be claimed by one service).

The -grpc service is only referenced as an internal Ingress backend (see
templates/ingress.yaml), so it never needs to be NodePort/LoadBalancer. This
sets it to ClusterIP and drops the nodePort field from its ports, which also
avoids a redundant second LoadBalancer when service.type: LoadBalancer.

Validation (helm template, chart pmm-1.8.1):

• Default (ingress disabled): only monitoring-service rendered, unchanged.
• NodePort + explicit nodePort + ingress enabled: monitoring-service keeps its
  nodePorts; monitoring-service-grpc renders as ClusterIP with no nodePort — no conflict.
• service.type=ClusterIP + ingress enabled: both ClusterIP, unchanged.
• helm lint and ct lint pass.

closes: #723

---

Drafted with Claude Code (Opus 4.8) and manually reviewed.

[it-percona-cla]

All committers have signed the CLA.

[somaz94]

Gentle ping 🙏 — green and ready for review. This sets the PMM gRPC service to ClusterIP to avoid a duplicate NodePort allocation. Happy to address any feedback. Thanks!

[somaz94]

Friendly ping on this one — it's a small fix for the duplicate nodePort conflict on the PMM gRPC service, and CI is green. Anything I can adjust to help it get a review? Thanks!