Support looking up rhcs_cluster_rosa_hcp data source by cluster name (not just id)

[lindleywhite]

Problem

The rhcs_cluster_rosa_hcp data source only accepts id (a Required string) as its lookup key:

data "rhcs_cluster_rosa_hcp" "cluster" {
  id = "cluster-id-123"
}

There is no way to look up a cluster by name. This is a real gap versus the underlying OCM API, which already supports name-based search — rosa describe cluster -c <name> and the raw clusters_mgmt REST API (GET /api/clusters_mgmt/v1/clusters?search=name+is+'<name>') both resolve a cluster by name today. I verified this directly:

curl -s -H "Authorization: Bearer $TOKEN" \
  "https://api.openshift.com/api/clusters_mgmt/v1/clusters?search=name%20is%20'imp-int-use2'"

returns the full cluster object (including id, aws.sts.oidc_endpoint_url, etc.) keyed only by name.

The underlying ocm-sdk-go client used by the provider also already exposes this capability:

clusterCollection.List().Search("name is '<name>'").SendContext(ctx)

(clusters_client.go — ClustersListRequest.Search(value string)).

Why this matters

In multi-workspace Terraform setups, a common pattern is:

1. Workspace A creates the ROSA HCP cluster (rhcs_cluster_rosa_hcp resource) and knows the cluster name at plan time (it's a variable), but the id only exists after creation.
2. Workspace B needs to configure the cluster after the fact (identity providers, group membership, etc.) and needs the id / sts.oidc_endpoint_url to reference it.

Today, the only ways to bridge this gap are:

• A terraform_remote_state cross-workspace lookup (many organizations intentionally avoid this pattern for state-isolation/security reasons), or
• Manually running rosa describe cluster -c <name> -o json | jq -r .id and hand-copying the id into workspace B's tfvars after every cluster (re)creation.

Since the cluster name is already known statically (it's a required input to the rhcs_cluster_rosa_hcp resource), letting the data source accept name as an alternate lookup key would let downstream workspaces resolve id (and other computed attributes like sts.oidc_endpoint_url) without remote state or manual copy/paste — using only a name-keyed data source, consistent with how the OCM API itself already works.

Proposed change

• Make id Optional + Computed instead of Required.
• Add name as Optional + Computed.
• Add a schema-level validator requiring exactly one of id / name to be set.
• In the data source's Read():
  • If id is set, keep the existing direct clusterCollection.Cluster(id).Get() path.
  • If only name is set, call clusterCollection.List().Search("name is '<name>'").SendContext(ctx), error if Total() != 1 (not found / ambiguous), and use the resulting cluster's id to populate the rest of the state (same as today).

Happy to submit a PR implementing this if the change is welcome — wanted to open the issue first per the contributing guide before submitting one.