Remove preStop hook that calls consul leave from Consul servers

[ishustava]

Background

Currently, when Consul servers are restarted, we call the consul leave command in the preStop hook that is registered with the servers. This approach has a couple of problems:

1. This command does not work when ACLs are enabled on the cluster because this command requires a valid ACL token with agent:write permission. Furthermore, we cannot reasonably make it work because even if the server's ACL token had that permission, we could not retrieve it because it's stored in the Consul's data directory. We could work around this by removing the preStop hook and instead setting leave_on_terminate to true on the servers, which would accomplish the same behavior. However, this solution is not recommended because of point 2.
2. The voting members of the consul cluster (aka the consul servers) should not gracefully leave the cluster. That is so that the quorum size is not affected during upgrades or downtime. See Leader still elected when below quorum consul#5947.

Changes proposed in this PR:

• Remove the preStop hook calling consul leave from the consul server pods

How I've tested this PR:

• Deployed Consul from this helm chart, ran kubectl rollout restart sts/..., and checked that servers came up healthy.
• Deployed the current stable version of the helm chart, upgraded to this version and checked that the upgrade goes through without problems

How I expect reviewers to test this PR:

Code review

Checklist:

• CHANGELOG entry added (HashiCorp engineers only, community PRs should not add a changelog entry)

[kschoche]

Great work @ishustava !
I'd personally call it a bug fix but all looks great, thanks for taking this on.

[ishustava]

I'd personally call it a bug fix but all looks great, thanks for taking this on.

good point, @kschoche. Let me fix that.

[ndhanushkodi]

🎉 looks great!