metrics(ci): create a metrics deployment specifically for CI smoke tests

[Maddiaa0]

Overview

When we run the smoke cluster and it fails, we are running without a metrics deployment, so we lose logging.

We should have a second metrics deployment that collects logs from the ci jobs.

• we can use some ci secret to authenticate access

[ludamad]

My idea for this was quite simple, have additional GitHub action jobs that just echo the logs

[ludamad]

We can already run on the same runner then it's just kubectl

[ludamad]

Actually even simpler, we can just have a bg process in the ci job that echoes the logs while the test runs