refactor: enhance restart command to utilize NodeProvider

[akrem-chabchoub]

Refactor: Use NodeProvider interface for restart command

Problem

The restart command was tightly coupled to the orchestration package, making it difficult to reuse and maintain. It had separate methods for cluster and namespace-based restarts with duplicated logic.

Solution

• Refactored to use NodeProvider interface for unified node discovery
• Simplified restart client with single Restart() method
• Added node group filtering capabilities
• Added direct pod restart support for namespace-based operations
• Improved error handling and logging

Changes

• Restart Client Refactoring: Replaced orchestration dependency with NodeProvider interface
• Node Group Filtering: Added support for filtering nodes by groups in beekeeper deployments
• Command Simplification: Unified cluster and namespace restart logic
• Pod-Level Restart: Added direct pod restart using label selectors for namespace operations
• Deployment Type Support: Added --deployment-type flag for beekeeper vs helm deployments

Usage Examples

# Restart with new image
beekeeper restart --cluster-name=my-cluster --image=bee:v2.0.0

# Namespace-based restart with pod label selector
beekeeper restart --namespace=my-namespace --label-selector="type:type1"

Closes #526

[gacevicljubisa]

c.setupCluster(ctx, clusterName, false) is already called in c.createNodeClient(ctx, true), no need to call it twice.

[gacevicljubisa]

If image is not specified, pod should only be restarted. If specified, then update the Image of the statefull set and then pod should be deleted if strategy is UpdateStrategyOnDelete

[gacevicljubisa]

If this method is enough to restart all bee nodes, we should rename it to Restart(), and remove RestartCluster()

[gacevicljubisa]

You do not need this logic. Much simplier and more precise solution would be to add NodeGroupName in bee.ClientOptions, and to expose new method NodeGroup() in bee package. Then you can compare on name. This should be done only in case of beekeeper deployment.

[gacevicljubisa]

This is psudo code how this filtering could work:

	filteredClients := make(map[string]*bee.Client)
	if len(sc.nodeGroups) > 0 && sc.deploymentType == DeploymentTypeBeekeeper {
		for _, ng := range sc.nodeGroups {
			for name, client := range sc.beeClients {
				if client.NodeGroup() == ng {
					filteredClients[name] = client
				}
			}
		}
	} else {
		filteredClients = sc.beeClients
	}

[akrem-chabchoub]

#516 (comment) , Its already implemented here:

beekeeper/pkg/restart/restart.go

Line 69 in ae823b3

func (c *Client) updateOrDeleteNode(ctx context.Context, node, ns, image string) error {