Fix Go Testing Pipeline

[rappm]

This pull request includes changes to the GitHub Actions workflows to improve the testing process and ensure better integration between jobs. The most important changes include adding a new test job, modifying the trigger for the Go test workflow, and enhancing the Go test command.

Improvements to Go testing workflow:

• .github/workflows/go-tests.yml: Updated the go test command to include the -race flag for detecting possible race conditions
• .github/workflows/go-tests.yml: Changed the trigger from pull_request to workflow_call to allow the workflow to be able to call it from the dev workflow
• .github/workflows/dev.yml: Added a new test-server job that uses the go-tests.yml workflow and set the build-and-push-server job to depend on the test-server job

[niclasheun]

The dev.yml is the deployment workflow.
This shall be independent of the testing workflow and not depend on each other. So please change this back and let the testing be a separate workflow.

[niclasheun]

(Updated Feedback - sorry)

I honestly do not see how the race flag actually resolve the root cause of the issue.
{"Time":"2025-02-03T10:54:40.078360835Z","Action":"output","Package":"github.com/niclasheun/prompt2.0/student","Test":"TestRouterTestSuite","Output":"2025/02/03 10:54:40 Failed to set up test database: failed to run SQL dump: failed to connect to user=testuser database=prompt:\n"}
{"Time":"2025-02-03T10:54:40.078401362Z","Action":"output","Package":"github.com/niclasheun/prompt2.0/student","Test":"TestRouterTestSuite","Output":"\t127.0.0.1:32780 (localhost): failed to receive message: unexpected EOF\n"}
{"Time":"2025-02-03T10:54:40.078410308Z","Action":"output","Package":"github.com/niclasheun/prompt2.0/student","Test":"TestRouterTestSuite","Output":"\t[::1]:32780 (localhost): failed to receive message: read tcp [::1]:42826-\u003e[::1]:32780: read: connection reset by peer\n"}

Hence, the issue is that Github cannot handle the amount of created test containers or that i.e. the timeout is set too short or something like that.

Race will only check the local storage of Go but not the connections of external containers. I assume it "fixes" the issue, because it introduces overhead and slows down the execution. But I do no see a reason how this actually resolves the root cause.

My Advice:

• look into the creation of the test containers in server/testutils
• refer to the test results file generated and uploaded by the workflow