fix: return error immediately on failed reconciliation status

[Aman-Cool]

Flux reconcile does not fail fast on Failed status

flux reconcile does not fail fast when a resource is already in Failed status.

The polling condition treats FailedStatus the same as InProgressStatus, so the command waits until the full timeout (default 5 minutes) before reporting an error.

This causes unnecessary delays and poor UX. The fix is to return an error for terminal failure states, matching the behavior of isObjectReady().

[Aman-Cool]

@matheuscscp @stefanprodan ,This makes flux reconcile fail fast when a resource is already in Failed status, instead of waiting for the full timeout. It aligns the behavior with isObjectReady() and improves CLI UX without affecting successful reconciliations.

[matheuscscp]

LGTM

[matheuscscp]

Please squash

[Aman-Cool]

@matheuscscp , Squashed into a single commit. All checks are green.

[Aman-Cool]

@stefanprodan ,Please see it when you have time, happy to help if it needs any updates. Thanks!

[themavik]

reconciliationHandled now returns fmt.Errorf(result.Message) for non-Current kstatus instead of (false, nil), so failed reconciles bubble up. nit: InProgress still returns false, nil—still the right shape for the wait loop?

[stefanprodan]

With the health check cancellation we've shipped in 2.8 this change would make things worse. We should be waiting and not bailout if Ready=False but the reason is https://github.com/fluxcd/pkg/blob/357bbcc3462ccf78cc17fd0b9b7d498b1848bf14/apis/meta/conditions.go#L162