failed to close job container stdout error

I've started to see the following in the containerd logs with

I0310 23:33:36.079002 4956 kuberuntime_manager.go:240] "Container runtime initialized" containerRuntime="containerd" version="v1.6.15"

{"ID":3,"EventName":"LogrusEntry","TraceEventProcessID":3400,"TraceEventProcessName":"","Message":"Cmd IO relay complete","Time":"2023-03-10T23:48:58.6121149+00:00","bytes":82,"eid":"e90354b911d2167a298c26bf7f9f47144e004a85efe2e64047bdc7f85c52f132","file":"stderr","pid":1404,"spanID":"0d238c510f703e97","tid":"17758b21e8e9ee7e7f22a9447d55f47c3d4258c2bcf30bc8009679716ba53f10","traceID":"cea6a1b58cd99d6db1a3d6c85bd8d9b1","error":"read |0: file already closed"}}

{"ID":4,"EventName":"LogrusEntry","TraceEventProcessID":3400,"TraceEventProcessName":"","Message":"failed to close Cmd stdout","Time":"2023-03-10T23:45:58.5958979+00:00","eid":"70dec87cbc63adbbc2c03a33cc4622a9ca77c1ce7ef02a09f926cb9972228f3e","pid":376,"spanID":"5127a9a44a063348","tid":"17758b21e8e9ee7e7f22a9447d55f47c3d4258c2bcf30bc8009679716ba53f10","traceID":"3138dc3e05b0a21de13cd39bf82104b8","error":"failed to close job container stdout: close |0: file already closed"}}

example: https://storage.googleapis.com/kubernetes-jenkins/pr-logs/pull/kubernetes-sigs_cluster-api-provider-azure/3273/pull-cluster-api-provider-azure-e2e/1634332087173517312/artifacts/clusters/capz-e2e-ankuc9-ha/kube-system/containerd-logger-l667r/containerd-logger.log

It doesn't seem to be causing an issue but makes the logs noisy.

Start by tracing the reported "failed to close Cmd stdout" and "Cmd IO relay complete" messages, using kuberuntime_manager.go:240 and the linked containerd-logger.log as entry points. Determine why an already-closed stdout is reported and define completion as eliminating benign log noise without hiding genuine close failures.