standarization of the registry domain config value for catalogs+apps

[piontec]

We use many different values in the values.yaml file of our helm charts to indicate the registry DNS domain that the images should come from. We have to come up with a single standard configuration layout that has to be implemented by all apps.

Lack of this standard is a problem, because:

• there's no single way to do it. Currently, we identified at lest 6 different approaches. Now, when we want to switch the registry domain globally by using the catalogs' config, there's a lot of duplication.
• some apps (*-cluster-apps) use strict values.yaml schema validation. Until we have a standard about it, there's again a ton of duplication we have to introduce to make them work (we push all the values to all the apps) and still some app's config can break it

[piontec]

blocked by #3185

[nprokopic]

some apps (*-cluster-apps) use strict values.yaml schema validation. Until we have a standard about it, there's again a ton of duplication we have to introduce to make them work (we push all the values to all the apps) and still some app's config can break it

Here if you refer to cluster-<provider> and default-apps-<provider>, see my comment here #3187 (comment)

TL;DR We should make a clear distinction between configuration of "regular" apps (that deploy controllers/operators) and cluster-<provider> (and default-apps-<provider>) apps. They are used for different things and "global" configs (changed "globally") have quite different impacts. Universal config for "regular" apps (that deploy controllers/operators) should IMO not be universal for and shared with cluster-<provider> (and default-apps-<provider>) apps as well.

[weatherhog]

waiting for https://github.com/giantswarm/giantswarm/issues/31101