feat: replace redis-cluster chart with redis-ha

[ferruhcihan]

📌 Summary

This PR replaces bitnami redis-cluster chart with redis-ha (https://github.com/DandyDeveloper/charts/tree/master/charts/redis-ha)

Ticket: https://track.akamai.com/jira/browse/APL-1079
PRs: apl-api | apl-docs

🔍 Reviewer Notes

Testing:

• Create a cluster (select g6-dedicated-8 to avoid hitting resource limits),
  or update the otomi-api app _rawValues in your cluster:

        image:
            # registry: docker.io
            # repository: linode/apl-api
            pullPolicy: Always
            tag: APL-1079

• In team-admin, navigate to the Catalog page in the console and add a new redis-ha Helm chart using this Chart.yaml link: https://github.com/linode/apl-charts/blob/APL-1079/redis-ha/Chart.yaml
• Go back to your desired team (e.g. team-demo)
• Follow the Network Policies lab with these changes.
• Verify that the new Redis pods and resources are working as expected.

[merll]

During startup I see the following of our provided Kyverno policies fail, where we should aim to eliminate as many as possible:

• require-pod-readiness-probe/validate-readiness-probe
• require-limits/validate-resources
• allowed-image-repos/good-repos
• require-pod-readiness-probe/validate-readiness-probe
• require-pod-startup-probe/validate-startup-probe
• require-non-root-groups/check-runasgroup
• require-non-root-groups/check-supplementalgroups
• require-non-root-groups/check-fsgroup

One more thing I wonder if we should reconsider, since we are making a change here anyway: Currently we sometimes attach the product name to the service, which especially in test scenarios can lead to awkward naming like redis-redis-ha. We are doing the same in RabbitMQ. For CloudnativePG on the other hand, we don't do this, so if one creates a database called db, the service name is db-rw or db-r which is fine. And since workload names are unique, it should be alright just to have the user decide what the service should be called.

So my suggestion is to name the service just like the workload, e.g. the user calls it cache, the service is called cache and not cache-redis-ha. Unless there is some other service that a workload could connect to, e.g. for read-only access.

[ElderMatt]

restore.ssh.image.tag needs to be encapsulated with ""