Setup CI/GitOps for cluster infra resources

[bwplotka]

I propose we apply infra resources to our GKE testing cluster on merge. There is a risk of "incidents", but for benchmarking cluster I would accept the risk.

Otherwise we need manual apply for certain PRs, which is arguably less reliable.

We could also setup a staging cluster or/and other validation for our manifests.

Acceptance Criteria

• Changes to prombench are tested and applied to production sooner and in more reliable way.

[metalmatze]

At Polar Signals we have a CI that takes all manifests that will be applied and kubectl diffs these against the actual cluster. The diff is then commented on the PR so it's crystal clear what changed.
I can take a look at how easy it would be to replicate for this environment.