[ci] Evaluate what to run on hosted and selfhosted runners

[alfi0812]

Evaluate what we want to run on our selfhosted runners and what to leave to github runners. Especially small tasks which no runtime should be kept on github runners

[Crow-Control]

In short:
The long-term goal of both building our own CI tools and self hosted runners, is to be less reliant on GitHub or other specific vendor-locked-in service providers.

Our runners are relatively well hardened and, for example, have no access to privileged components such as docker-daemons.

This also means that certain tasks, currently, inherently cannot run on self-hosted runners:

• containerforge: container build and test tasks
• truecharts: tasks relying on k3d/kind: chart-tests and common-tests (specifically the install-tests).

Especially small tasks which no runtime should be kept on github runners
The size of the run shouldn't matter.
However, in line with the goal of moving away from GitHub-actions specific logic with time, we should look into ensuring a few number of big jobs. Preferably sorted on the runner kind they need to run on.

At the same time, we should slowly move away from pre-made GitHub Actions, to our own logic preferably included in our runner container and/or forgetool. This should also limit the amount of needless load these things put on the runners.