Consider using `make test-ci` instead of `make test-image` for unit test pipelines

[janbrasna]

Description

While investigating #16495 it turned out the feature I was after would get covered in a release build that's part of make test-ci — however the unit tests only build dev/test containers in a different order using make test-image skipping the release service.

There might be some tribal knowledge how these two differ, or apparent differences when each target command is appropriate, so I'll keep investigating to get the idea, but would welcome any insight into these two, and the viability of the idea to switch to the different service set to cover more prod-like build context to catch #16495 with it.

---

Success Criteria

• Decide on the correct make target to run in CI
• Make sure there's no gap between the two targets that would cover different surface or cases
• Use the additional release spec service to make sure the strict image build for deployment later has a higher chance of succeeding

[janbrasna]

So the test run should be the same really, as after untangling the targets and commands (and ignoring some extra env getting set), the two basically run:

make test-image:

1. DC build app assets
2. DC run test-image

make test-ci:

1. DC build --pull release
2. DC build app builder assets app-base
3. DC run test-image

The latter, being pretty much what's run later in CD, also checks for missing static files referenced addressing #16495 — and comparing a handful of runs, taking approx. extra 60–90sec in the CI in general.

(I'm currently investigating layer caching between runs that's not being actually used at this moment, which could speed this up significantly — using the time saved for more thorough checks.)