Toggable cloud-init application verbose option

[anmazzotti]

Problem:
As a user I would like to be able to debug failed cloud-init applications through logs.

There are some instances where the runcmd commands fail and at the same time are not idempotent, so manually repeating the commands a next time to debug may produce different results from the original failure.

Another instance where logs are required is when the runcmd list contains destructive commands, so that in case any of those fails it will be hard to reproduce the same scenario again and debug or retrieve information about the root cause of the failure.

Desired solution:
I'd like to configure elemental-toolkit to be able to print all runcmd execution logs and potentially to print verbose information for the all cloud-init configs application, so that failures can be analyzed from logs.

Ideally this should be configurable on a host level, for example using a global /etc/elemental/conf.yaml config.

The verbose option could also implemented per-config, so that I can decide one cloud-init application should be more verbose than another one, but not a requirement for this issue, just an idea.

[anmazzotti]

@kkaempf this is mostly CAPI related as the elemental toolkit does the k8s bootstrapping.
For the current elemental-operator the rancher-system-agent does it and iirc it does already print verbose logs.

The question being if this should have slem6 milestone or not.