Investigate the feasibility/advantages of using pulp-operator in pulp-container CI

[git-hyagi]

Right now, we cannot run functional tests for replication because we don't have a pipeline job running with multiple instances of Pulp.
In pulp-operator project we have a workflow to test multiple installations of Pulp (actually we are deploying only 2 because of the limited resources):
https://github.com/pulp/pulp-operator/actions/runs/9410984818/job/25923537636#logs
We could investigate if it is worth reusing the current pulp-operator pipeline in pulp_container to run tests that would need multiple running Pulps.

Things to consider:

• the current pulp-operator pipeline (which are only installing 2 instances of Pulp) is taking 10 minutes to run
• this new pipeline would need a new step to rebuild the image to test the PR changes and update the operator to run with this new image
• we don't know if GH runners will have enough resources (CPU/memory/disk) to handle the k8s cluster + pulp-operator + 2xPulp + replication workload

[lubosmj]

Does it take 10 minutes to run the setup (like we have here install.sh) or together with tests (if there are any)? Currently, the initialization in our case (pulp_container's CI) takes roughly 5 minutes.

[git-hyagi]

10 minutes to run everything (it takes around 8 minutes just waiting for Pulp to run all migrations and the other ~2 minutes are spent installing k8s, OLM, pulp-operator, etc.).

[mdellweg]

We have sucessfully created tests to replicate from one domain to another in the same pulp container. Obviously this can never test replicating from different plugin versions...