A bunch of e2e test improvements, such us:

• pause_scaledjob_test.go is a flaky test because it's based on passed times during specific moments and it can depend on cluster state. This PR changes the test adding different checks to validate the different statuses, not depending on checking the things in the exact moment.
• scaling_modifiers_test.go doesn't validate that the mock metrics server is ready before starting the tests, producing errors because the test starts without the mocks ready
• pentelemetry test doesn't validate the lag metric values as it's not relevant for the test, but it introduces a not controlled variable which makes the test fail
• rabbitmq helper now waits until the server is ready
• etcd scaler uses helm chart for deploying etcd
• tests fail fast if there is any issue with dependencies (e.g: if Kafka cluster doesn't start, we fail the tests directly instead of waiting until the end)

TODOs:

• There is an issue in rabbitmq server which produces failures with oauth. The issue is fixed and the release was cut, but we have to wait until docker releases the version too. In the meantime I have just downgraded the version, but we should use the fixed one once it's released

Checklist

• Commits are signed with Developer Certificate of Origin (DCO - learn more)

Fixes #5547