### Overview The `PHPUnit-Kernel` test suite in `.gitlab-ci.yml` runs across 3 parallel CI jobs (`CI_PARALLEL_NODE_TOTAL: '3'`). As the kernel suite has grown, each shard takes longer, increasing pipeline wall-clock time. Splitting the suite across 4 jobs instead of 3 reduces per-shard runtime. This was first done as a drive-by change inside MR !1271 (commit d6cfd28e, by Wim Leers), but it is a standalone CI-capacity change with no dependency on that issue's feature work, so it is split out here. ### Proposed resolution In `.gitlab-ci.yml`, change the `PHPUnit-Kernel` matrix entry to spread across 4 parallel jobs: - `CI_PARALLEL_NODE_INDEX: ['1', '2', '3']` -> `['1', '2', '3', '4']` - `CI_PARALLEL_NODE_TOTAL: '3'` -> `'4'` Confirm the existing `KUBERNETES_CPU_REQUEST` and `DRUPAL_TEST_CONCURRENCY` values still match Drupal core's configuration for the kernel suite. ### Remaining tasks - Update the `PHPUnit-Kernel` parallelism in `.gitlab-ci.yml` from 3 to 4 jobs. - Verify pipeline wall-clock time improves and shards stay balanced. ### User interface changes None in this issue. ### API changes None. ### Data model changes None.