ci: Connect Lighthouse CI to database on dev

[ZeroWave022]

The lighthouse CI GitHub Actions workflow is now connected to the database running on our dev server.
I have not found any solution which could "reuse" our already running build on the dev server, but if you manage to find something that could work, feel free to change the PR.
The workflow runs only on pull requests, so it runs the absolute minimum amount of times, therefore consuming less of our minutes available for workflows.

[ZeroWave022]

When the Docker configuration was changed from using volumes to referring to actual folder paths on the server, the data we had earlier has been effectively nuked. The secrets are updated, and our LHCI will start gathering new data from now on.

[michaelbrusegard]

Good work. I had some questions before approving.

1. How is it able to connect to the database on noodlebar? You are only using the local docker name for the database so I dont know how it works.
2. We also have to setup it to connect to the local s3 server the same way so it can load images.
3. You can remove all the FEIDE_ and MATRIX_ environment variables as those have been set to optional.
4. Can we fix it so it stops failing, why is it failing?

[michaelbrusegard]

Good work. I had some questions before approving.

1. How is it able to connect to the database on noodlebar? You are only using the local docker name for the database so I dont know how it works.

2. We also have to setup it to connect to the local s3 server the same way so it can load images.

3. You  can remove all the FEIDE_ and MATRIX_ environment variables as those have been set to optional.

4. Can we fix it so it stops failing, why is it failing?

I will create a separate issue for fixing the lighthouse issues. Some of the ones on the auth page are quite useful, like why is the title missing? Also we can then create a CSP on that same branch

[ZeroWave022]

Right, I am dumb. I assume we dont want to expose the database, so this needs to run on noodlebar so it has access to the services. Will need to move it to the self-hosted runner there. Hopefully, there's an option to force using the dev server and not anything else. I think GH actions is set up there already?
It will continue failing until there are no errors reported from Lighthouse. Warnings are fine as far as I know.
I'll update the config to reflect the changes

[ZeroWave022]

Lighthouse CI is the least documented tool I've ever interacted with. After finding what environment variable needs to be set by looking at the source code, LHCI can't connect to Chrome. I'm not looking anymore into this, if you want to look at it, feel free to