sylvactl watch goes back to previously reconciled units (calico-ready, misc-controllers)

In https://gitlab.com/sylva-projects/sylva-core/-/jobs/13008827647, we see the following:

• (keeping in mind that cluster unit depends on misc-controllers-suite unit)

• cluster unit starts reconciling at 2026/02/05 23:41:59.170396 Kustomization/cluster state changed: Progressing

• at multiple times during the run the following sequence occurs:

  • the misc-controllers-suite goes into health check failed

    2026/02/05 23:54:15.068568 Kustomization/misc-controllers-suite state changed: HealthCheckFailed - health check failed after 30.016270739s: timeout waiting for: [Deployment/misc-controllers-suite-system/misc-controllers-suite-controller-manager status: 'InProgress']

  • sylvactl starts to wait for it instead of cluster

    2026/02/05 23:54:15.072409  ╭╴Waiting for the following resources to progress:
    2026/02/05 23:54:15.072418  │  Kustomization/misc-controllers-suite               HealthCheckFailed - health check failed after 30.016270739s: timeout waiting for: [Deployment/misc-controllers-suite-system/misc-controllers-suite-controller-manager status: 'InProgress']
    2026/02/05 23:54:15.072421  ╰╴╴╴┄        

  • some times after, misc-controllers-suite recovers, and sylvactl watch goes back to cluster unit

    2026/02/05 23:55:15.420725 Kustomization/misc-controllers-suite state changed: ReconciliationSucceeded - Applied revision: 0.0.0-git-d2d089e1@sha256:1e3f3496c99ea01a7569311284abd25907edd903750105da95bc4389614f2861
    2026/02/05 23:55:15.424543  ╭╴Waiting for the following resources to progress:
    2026/02/05 23:55:15.424555  │  Kustomization/cluster               DependencyNotReady - dependency 'sylva-system/cluster-node-provider-id-blacklist' is not ready
    2026/02/05 23:55:15.424557  ╰╴╴╴┄

• the last times this occurs, the misc-controllers-suite healthcheck is still failing while sylvactl global timeout triggers, and sylvactl watch is only waiting for misc-controllers-suite, and hence gives this message:

2026/02/06 01:20:48.730985 Kustomization/misc-controllers-suite state changed: HealthCheckFailed - health check failed after 30.020314463s: timeout waiting for: [Deployment/misc-controllers-suite-system/misc-controllers-suite-controller-manager status: 'InProgress']
2026/02/06 01:20:48.735663  ╭╴Waiting for the following resources to progress:
2026/02/06 01:20:48.735677  │  Kustomization/misc-controllers-suite               HealthCheckFailed - health check failed after 30.020314463s: timeout waiting for: [Deployment/misc-controllers-suite-system/misc-controllers-suite-controller-manager status: 'InProgress']
2026/02/06 01:20:48.735682  ╰╴╴╴┄
2026/02/06 01:21:48.755469 Kustomization/misc-controllers-suite state changed: Progressing - Reconciliation in progress
2026/02/06 01:22:18.018452 Command timeout exceeded: Kustomization/sylva-system/sylva-units-status did not became Ready after 2h7m0s

<inventory dump for Kustomization/misc-controllers-suite>

When sylvactl watch has seen a Kustomization become Ready, it is not supposed to wait again for it later in the same run, but here in this example, it does.

We've already seen this occur with calico-ready unit (on which cluster unit also depends-on).