Devops: Move RabbitMQ CI to nightly and update versions (#6419)

The `rabbitmq.yml` was running a subset of the unit test suite against
a number of versions of RabbitMQ. The versions tested were very outdated
and are not officially supported anymore. The versions are updated to
3.11, 3.12, and 3.13.

Since the introduction, the tests have never really failed, which is not
too suprising as the code in `aiida-core` doesn't actually directly
interface with RabbitMQ. Rather this is done in `kiwipy`. Therefore, the
tests are moved to the nightly workflow such that they are no longer
run on every push and PR, which is overkill.

.github/workflows/nightly.yml

@@ -21,7 +21,7 @@ concurrency:
 
 jobs:
 
-  tests:
+  nightly-tests:
 
     if: github.repository == 'aiidateam/aiida-core'      # Prevent running the builds on forks as well
     runs-on: ubuntu-latest
@@ -76,9 +76,9 @@ jobs:
       run: .github/workflows/tests_nightly.sh
 
     - name: Slack notification
-            # Always run this step (otherwise it would be skipped if any of the previous steps fail) but only if the
-            # `install` or `tests` steps failed, and the `SLACK_WEBHOOK` is available. The latter is not the case for
-            # pull requests that come from forks. This is a limitation of secrets on GHA
+      # Always run this step (otherwise it would be skipped if any of the previous steps fail) but only if the
+      # `install` or `tests` steps failed, and the `SLACK_WEBHOOK` is available. The latter is not the case for
+      # pull requests that come from forks. This is a limitation of secrets on GHA
       if: always() && (steps.install.outcome == 'Failure' || steps.tests.outcome == 'Failure') && env.SLACK_WEBHOOK != null
       uses: rtCamp/action-slack-notify@v2
       env:
@@ -88,3 +88,63 @@ jobs:
         SLACK_COLOR: b60205
         SLACK_TITLE: Nightly build of `aiida-core/main` failed
         SLACK_MESSAGE: The tests of the `nightly.yml` GHA worklow failed.
+
+
+  # Run a subset of test suite to ensure compatibility with latest RabbitMQ releases
+  rabbitmq_tests:
+
+    runs-on: ubuntu-latest
+    timeout-minutes: 30
+
+    strategy:
+      fail-fast: false
+      matrix:
+        # Currently supported RMQ versions per:
+        # https://www.rabbitmq.com/docs/which-erlang#compatibility-matrix
+        rabbitmq-version: ['3.11', '3.12', '3.13']
+
+    services:
+      postgres:
+        image: postgres:16
+        env:
+          POSTGRES_DB: test_aiida
+          POSTGRES_PASSWORD: ''
+          POSTGRES_HOST_AUTH_METHOD: trust
+        options: >-
+          --health-cmd pg_isready
+          --health-interval 10s
+          --health-timeout 5s
+          --health-retries 5
+        ports:
+        - 5432:5432
+      rabbitmq:
+        image: rabbitmq:${{ matrix.rabbitmq-version }}-management
+        ports:
+        - 5672:5672
+        - 15672:15672
+
+    steps:
+    - uses: actions/checkout@v4
+
+    - name: Install aiida-core
+      uses: ./.github/actions/install-aiida-core
+
+    - name: Install system dependencies
+      run: sudo apt update && sudo apt install postgresql
+
+    - name: Run tests
+      run: pytest -sv -k 'requires_rmq'
+
+    - name: Slack notification
+      # Always run this step (otherwise it would be skipped if any of the previous steps fail) but only if the
+      # `install` or `tests` steps failed, and the `SLACK_WEBHOOK` is available. The latter is not the case for
+      # pull requests that come from forks. This is a limitation of secrets on GHA
+      if: always() && (steps.install.outcome == 'Failure' || steps.tests.outcome == 'Failure') && env.SLACK_WEBHOOK != null
+      uses: rtCamp/action-slack-notify@v2
+      env:
+        SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}
+        SLACK_ICON: https://www.materialscloud.org/discover/images/0ba0a17d.aiida-logo-128.png
+        SLACK_CHANNEL: dev-aiida-core
+        SLACK_COLOR: b60205
+        SLACK_TITLE: RabbitMQ nightly tests of `aiida-core/main` failed
+        SLACK_MESSAGE: The rabbitmq tests in the `nightly.yml` GHA worklow failed.