Nice!
Can you please also address as part of this PR:

• [Terraform] Model referencing is broken between cos-lite module and charm modules observability#344
• As part of CI, we should terraform apply the minimal cos and cos-lite reproducers and juju wait-for active idle.

The COS reproducer won't go active/idle because it doesn't include s3 storage. Arguably, that change is more involved than we need to.

I also imagine that if this CI ran on PRs, it would gate small reproducer-unrelated changes behind a very long test.

Personally, I would keep the CI changes to a separate PR.

I think if we add a terraform validate step to our PR workflow, this would catch most regression bugs and general module integrity

FYI issues are syncing to OPENG, I don't remember what we said about ownership for these issues. For example:

• Cleanup the COS Lite terraform module to match spec / standard #37

Will we just remember that they exist there and create new ones in O11y-stack?

Merging on failing docs and TF CI because the migration caused mismatches. This will be resolved once the merge takes place. Tested with tfv and terraform fmt. The docs are pointing at the main branch which currently does not have the latest changes.