bug: Retrieve the workspace-related CoreDNS IP from the Service resource instead of using DNS lookup.

[davidfestal]

Describe the bug

Currently, when mutating a Deployment to override the DNS server, the Syncer tries to guess the IP of the CoreDNS service by issuing a DNSLookup on the internal service dns name.
This prevents using the Syncer process outside of the Synctarget cluster (for local debugging purposes for example).
A better and more robust way is to get the clusterIP on the Service resource.

Steps To Reproduce

1. Run the Syncer in it own local process
2. Observe that, when trying to sync a deployment, you would be the error:

E1115 18:44:16.121384  418586 spec_controller.go:295] kcp-workload-syncer-spec failed to sync {{"apps" "v1" "deployments"} "root:org:user|default/httpecho"}, err: failed to get DNS nameserver IP address: lookup kcp-dns-us-west1-1mwhl3gm-x0gx4iqu.kcp-syncer-us-west1-1mwhl3gm.svc.cluster.local: no such host

Expected Behaviour

We should still be able to run the syncer locally.

Additional Context

No response