negative nativeHistogramBucketLimit breaks reconciliation

Creating a ServiceMonitor with nativeHistogramBucketLimit set negative (outside range of uint64) can break the operator, it was not reconciling changes to Prometheus instances while a ServiceMonitor in the namespace had the negative value set.

COO Version: 1.3.1
OCP Version: 4.18.33

ts=2026-06-12T18:25:21.159784115Z level=error caller=/cachi2/output/deps/gomod/pkg/mod/k8s.io/client-go@v0.34.2/tools/cache/reflector.go:205 msg="Failed to watch" logger=UnhandledError err="failed to list *v1.ServiceMonitor: json: cannot unmarshal number -2 into Go struct field ServiceMonitorSpec.items.spec.NativeHistogramConfig.nativeHistogramBucketLimit of type uint64" reflector=pkg/mod/k8s.io/client-go@v0.34.2/tools/cache/reflector.go:290 type=*v1.ServiceMonitor

  apiVersion: monitoring.rhobs/v1
  kind: ServiceMonitor
  metadata:
...
  spec:
    nativeHistogramBucketLimit: -2

Start by tracing ServiceMonitorSpec.nativeHistogramBucketLimit through the ServiceMonitor watch and reconciliation path, using the negative-value manifest and reflector error as the reproduction. Verify that a negative nativeHistogramBucketLimit no longer prevents reconciliation of Prometheus instances, and add or run the relevant validation or reconciliation checks if they exist.