Stabilize Playwright Unit Tests by Centralizing Global Mocks

Input from Gemini 3.1 Pro:

✦ The root cause of the random test suite failures was isolated to an architectural race condition during test teardown, not a global mock pollution issue.

The Problem: In unit tests (like GridScrollProfile.spec.mjs), setting a config like dragResortable: true or responsive: true triggers an async module load via import() inside the respective afterSet... hook on container.Base or component.Base. When the test completes, it synchronously calls .destroy() on the component hierarchy, unregistering it from the ComponentManager. However, Playwright worker processes are reused. If the pending import() microtask resolves after the test has finished (while the worker is spinning up the next test file), the hook resumes execution against a now-destroyed instance. For afterSetDragResortable, it would then call me.createSortZone(), which attempts to access me.parent.id. Because me.parent dynamically queries the ComponentManager (where the instance no longer exists), it returned null, crashing the entire Playwright worker with a TypeError.

The Solution: The framework's built-in trap() method was designed precisely to handle this scenario by safely rejecting pending async operations when a component is destroyed. We wrapped the dynamic module imports in src/container/Base.mjs and src/component/Base.mjs with this.trap():

module = await me.trap(me.loadSortZoneModule());
// ...
module = await me.trap(import(`../../src/plugin/Responsive.mjs`));

This ensures that if the test suite tears down the component before the network/file system resolves the module, the execution gracefully aborts rather than blowing up the worker environment for subsequent tests.

The entire unit test suite now runs reliably in a single worker process without cross-contamination.