Achieving Even Distribution of Kubernetes N+1 Replicas Across N Nodes

Question

Suppose I have 3 nodes and 4 replicas of a deployment in Kubernetes. My objective is to ensure that, whenever possible, two replicas under the same deployment should not be scheduled on the same node.

For example, when dealing with 2 replicas and 2 nodes, I can apply pod anti-affinity to prevent the assignment of both pods to a single node:

      affinity:
        podAntiAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
          - labelSelector:
              matchExpressions:
              - key: app
                operator: In
                values:
                - frontend
            topologyKey: kubernetes.io/hostname

However, when there are 4 replicas, the issue arises where 3 pods are successfully scheduled on 3 distinct nodes, but 1 pod remains in a pending state due to the inability to satisfy the pod anti-affinity rules on a single node. Is there a method to ensure that each replica is allocated to a separate node if available, falling back to scheduling on an existing node if necessary?