migration job never properly finishes when using slq-proxy sidecar

[zaibon]

Since we added the sql-proxy to the migration pod. the pod stays alive forever even after the job is done.
It seems this particular case is not yet supported natively by Kubernetes. Alouthgh there is an open KEP for it: https://github.com/SergeyKanzhelev/enhancements/tree/ace2cc6431e0be029e15d4c2407b8fdac9ea765c/keps/sig-node/753-sidecar-containers. But it's not planned to be stable before v1.30.

This bug makes the helm chart installation to always be a failed state since it waits for the job to finish. We should find a way to at least allow Helm to consider the installation successful.
Then in a second time, we should find a solution to allow the migration pod to shutdown completely.