feat(cluster)!: use min_nodes instead of node_count

[adambabik]

Motivation

node_count is a static value which does not describe
the reality. ScyllaDB clusters are dynamic and
they scale out automatically when the disk is
running out of scape.

node_count also can't be used to resize the cluster
manually as any change to it results in recreation
of the resource.

This change introduces min_nodes and makes
node_count a computed property. min_nodes
issues a resize request and waits for the completion.

Note that there is no max_nodes as the platform
does not support it at the moment.

Summary

• This PR introduces a breaking change.
• node_count becomes a computed property.
• min_nodes controls the minimum number of nodes. It works this way that any change to min_nodes, after the resource is created, issues a resize request. The value is updated only if the resize request succeeds.
• Note that this change is only for the Standard clusters. In the future, Terraform Provider will support X Cloud where a scaling policy will be set and work to describe a desired state. The ScyllaDB Cloud internal services will monitor the cluster and resize it to conform to the desired state.

Todo

• Test the following scenario: create a cluster using TF, scale-out using API, refresh the TF state. Expectation: node_count should be updated but the plan should be empty. Next, update min_nodes to equal node_count. Expectation: min_nodes should be updated, but no resize should be triggered.
• Test the following scenario: create a cluster using TF with min_nodes = 6, scale-in using API, refresh the TF state. Expectations: node_count should be updated, the plan should result in resize.

[Copilot]

Pull request overview

This PR introduces a breaking schema change for the scylladbcloud_cluster resource to model cluster sizing dynamically: node_count becomes computed, while min_nodes becomes the user-controlled input that triggers resize operations.

Changes:

• Replaces node_count (static/ForceNew) with min_nodes (updatable) and makes node_count computed from actual active nodes.
• Adds resize support in the Scylla client and implements update behavior to issue a resize request and wait for completion.
• Updates acceptance tests and state migration (SchemaVersion 2) to support upgrading existing state from node_count to min_nodes.

Reviewed changes

Copilot reviewed 9 out of 10 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
internal/tfcontext/context.go	Removes provider endpoint logging helper; keeps HTTP request info fields.
internal/scylla/endpoints.go	Adds ResizeCluster and refactors ListClusterRequest to accept a params struct.
internal/provider/serverless/serverless_cluster.go	Updates cluster-request polling calls to new ListClusterRequest/wait helper naming.
internal/provider/provider_test.go	Updates/extends acceptance tests for min_nodes behavior and migration coverage.
internal/provider/cluster/cluster_v1.go	Adds schema v1 upgrader to migrate node_count → min_nodes.
internal/provider/cluster/cluster_v0.go	Moves v0 state upgrade logic into versioned file (adds disk size upgrade).
internal/provider/cluster/cluster.go	Implements SchemaVersion 2, adds min_nodes, adds resize-based Update, and introduces request-wait helpers.
go.mod / go.sum	Adds testify and updates indirect deps for new test helpers.
Makefile	Tweaks test timeouts/parallelism and adds a target to list acceptance tests.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.