fix: reduce noisy PodClique NotFound logs during cascade-delete

[AsadShahid04]

Summary

• Lower the `GetPodClique` not-found log from `Info` to `V(1)` (debug) so routine cascade-delete reconciles no longer spam the log at default verbosity
• Add a contextual `Info`-level log in the PCS gang-termination path (`getMinAvailableBreachedPCLQsNotInPCSG`) when a PodClique expected by a live PodCliqueSet replica is missing, preserving visibility for unexpected absences
• Add unit tests: rename the `not_found` test case to clarify it exercises the ignore-not-found path, and add a new `TestGetPodCliqueIgnoreNotFoundFalse` covering error propagation when `ignoreNotFound=false`

Problem

After the cascade-delete change in #556, deleting a PodCliqueSet removes the PCS and its owned PodCliques. The PodClique controller then receives reconcile events for already-deleted PodCliques and logs a info-level "PodClique not found" message for each one. At scale (e.g. 5000-replica workloads) this produces thousands of expected but noisy log entries during normal deletion.

Solution

operator/internal/controller/utils/reconciler.go — Change the GetPodClique not-found branch from logger.Info to logger.V(1).Info. Normal operators run at verbosity 0, so this message disappears from default logs entirely while remaining accessible with -v=1.

operator/internal/controller/podcliqueset/components/podcliquesetreplica/gangterminate.go — In getMinAvailableBreachedPCLQsNotInPCSG, the function already collects the names of expected PodCliques that are missing. Previously it silently skipped the replica's MinAvailable evaluation. Now it first emits an Info log naming the missing PodCliques and the affected replica index. This is the correct level here: the PCS is alive and not deleting, so a missing PodClique is a transient but observable state (e.g. creation still in flight), not a guaranteed cascade-delete.

Testing

• go build ./... — passed
• go test ./internal/... — all unit tests pass (no cluster required)
• New test TestGetPodCliqueIgnoreNotFoundFalse verifies error propagation; updated not_found_ignore case documents debug-level behaviour

Closes #622

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.