ci: always run all tests (#397)

### Summary of Changes

Running only the tests of changed components leads to incorrect coverage
reports and prevents us from making the tests required. The latter
allows PR with failing tests to be merged, like #391.

.github/workflows/pr.yml

@@ -9,31 +9,8 @@ concurrency:
   cancel-in-progress: true
 
 jobs:
-  # Run change detection
-  changes:
-    runs-on: ubuntu-latest
-    permissions:
-      pull-requests: read
-    outputs:
-      dsl: ${{ steps.filter.outputs.dsl }}
-      runner: ${{ steps.filter.outputs.runner }}
-      stdlib: ${{ steps.filter.outputs.stdlib }}
-    steps:
-      - uses: dorny/[email protected]
-        id: filter
-        with:
-          filters: |
-            dsl:
-              - 'DSL/**'
-            runner:
-              - 'Runtime/safe-ds-runner/**'
-            stdlib:
-              - 'Runtime/safe-ds/**'
-
   # Build and test DSL component
   build-dsl:
-    needs: changes
-    if: ${{ needs.changes.outputs.dsl == 'true' }}
     runs-on: ubuntu-latest
     defaults:
       run:
@@ -80,8 +57,6 @@ jobs:
 
   # Build and test Runtime > Runner component
   build-runtime-runner:
-    needs: changes
-    if: ${{ needs.changes.outputs.runner == 'true' }}
     runs-on: ubuntu-latest
     defaults:
       run:
@@ -132,8 +107,6 @@ jobs:
 
   # Build and test Runtime > Stdlib component
   build-runtime-stdlib:
-    needs: changes
-    if: ${{ needs.changes.outputs.stdlib == 'true' }}
     runs-on: ubuntu-latest
     defaults:
       run: