arche2025h1 configs

[phbnf]

Towards #104

This PR introduces arche2025h1 log config. It's not an end state, but reflects how I've been running it for some time now, so let's get this in.

Another PR will configure the preloader, I thought I'd split them in two for ease of reviewing. Some part of this current PR might mention the preloader, because it's coming in the PR just after.

This PR:

• introduces's arche2025h1 terraform and terragrunt configs: configs in this PR are based on pre-existing configs, with some minor differences that you can see in invidivual prs.
• defines arche2025h1's roots, and explains how to update them. We ought to be something better later, but it's okay for now.
• triggers a log deployment each time we tag a commit with "deploy-staging-XX" to make development faster for now

[roger2hk]

Add the static-ct-staging directory to deployment/live/gcp/README.md.

[roger2hk]

Is it likely to have another env under the static-ct-staging directory? Consider merging this file with deployment/live/gcp/static-ct-staging/cloudbuild/staging/terragrunt.hcl?

[phbnf]

Yes agreed that can be changed - I paused here, but I'm actively thinking about changing this, and maybe having a cloudbuild for the log, and a different one to run the preloader. I'll leave things like this for now, to make sure they mirror the static-ct setup.

[roger2hk]

Suggested change

	TODO(phboneff): come back to this, CloudRun doesn't trigger a deployment if the tag value change value.
	TODO(phboneff): come back to this, Cloud Run doesn't trigger a deployment if the tag value change value.

[roger2hk]

Do you mean the tag is re-tagged to another commit and force pushed?

[phbnf]

I couldn't parse my own sentence, sorry. If a Cloud Run service was started via terraform with a container tagged with "latest", a new (different!) container is tagged "latest", and the terraform config is applied again, the Cloud Run service won't restart. A fix is to use tags that are non stable across containers, such as their SHORT_HASH. It's not an issue for the ci stack, because we bring it down and up every time.