Improved handling of consumer checks for volume migration

[noctarius]

If the consumer is in pending, in-creation state, a volume migration is not allowed to start and has to wait for the pod to be running. If the pod is stopped (replica=0) or the pvc is unbound, we can skip the connect+validation step.

This PR improves the detection and ensures the coming-up pod isn't interrupted or corrupted. This is a follow up of #316.

[Copilot]

Pull request overview

Improves volume migration safety by (1) distinguishing “no consumer” vs “consumer not ready” when deciding whether to run NVMe validation, and (2) making the ContinueMigration step crash-safe by checking the backend migration phase before issuing the continue call (follow-up to #316).

Changes:

• Add uncached Kubernetes reads (APIReader) and richer consumer-state detection to avoid skipping validation for genuinely active volumes and to wait when a consumer pod is Pending.
• Introduce performMigration to make ContinueMigration idempotent by checking backend MigrationDTO.Phase before continuing.
• Expand unit test coverage for skipped validation, waiting behavior, and non-idempotent continue hazards.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 3 comments.

File	Description
internal/webapi/rebalancing.go	Adds migration phase constant and a helper to classify terminal migration statuses.
internal/controller/volumemigration_controller.go	Updates consumer detection (uncached reads + new error cases) and adds crash-safe performMigration logic around ContinueMigration.
internal/controller/volumemigration_controller_unit_test.go	Adds tests for consumer-state behavior and for safe ContinueMigration retry semantics.

---

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