feature: Handling migrating Workspaces from shard to shard

[ntnn]

Feature Description

Currently the provider.Provider watches a single VW and manages all clusters (in a mcr sense and logical cluster in kcp sense) for that shard.
When Workspace migration is implemented (kcp-dev/kcp#3498) this will operators cause to delete resources for workspaces being migrated because they will vanish from the VW of one shard and reappear in another.

That largely depends on how the migration is implemented in kcp and how transparently the move is communicated.

Proposed Solution

Depends on implementation in kcp.

Alternative Solutions

No response

Want to contribute?

• I would like to work on this issue.

Additional Context

No response

[ntnn]

With the refactor of pkg/provider we don't have the problem of the underlying provider stopping a cluster anymore.

But the wildcard cache will probably need to invalidate cached objects of migrating clusters.
The cluster config probably doesn't need to be touched as it should go through the init/term VW content proxy or through the front-proxy.