- when we forward the request to the aggregated server, set the audit
ID in the new request header. This allows audit logs from aggregated
apiservers to be correlated with the kube-apiserver.
- use the audit ID in the current tracer
- use the audit ID in httplog
- when a request panics, log an error with the audit ID.
Kubernetes-commit: b607ca1bf3e1cf6152c446ea61ac7fdd9014e1f1