* fix: Reap orphaned ("zombie") processes in argocd-repo-server pod

In openshift clusters, the user id of your container can be arbitrary,
so you need to make the running images compatible with this behavior.

The problematic application for argo-cd was the repo server. When trying
to clone the repos it was getting the error "No user exists for uid
100083000" (100083000 was the random user id being injected by
openshift in my case). This was because the user 100083000 wasn't in the
/etc/passwd file.

The changes in this commit add a uid_entrypoint.sh script that, when the
container starts, modifies the /etc/passwd file to add an entry with the
current UID _only_ if the current UID isn't there.

References:
* Problematic behavior of ssh when user id isn't in the /etc/passwd file:
  https://unix.stackexchange.com/questions/524268/running-git-or-ssh-client-in-docker-as-user-no-user-exists-for-uid
* OpenShift guidelines on how to make your docker image runnable by
  arbitrary user ids:
  https://access.re...