To optimize latency issues in a a set of collaborating pods within one project, we would like to request a "default" pod affinity configuration that a platform administrator can define per namespace (similar to the default node selector).

With this feature, we would be able to  bind a set of collaborating pods to a preferred Availability Zone (if a pod does not explicitly specify a pod affinity); this would be used to minimize cross site traffic, yes, with the price of reduced availability).

Related:
https://kubernetes.io/docs/concepts/configuration/assign-pod-node/#affinity-and-anti-affinity