fix(tests): use forks pool in CI to prevent flaky worker timeouts

Switched from threads pool to forks pool in CI environments to prevent intermittent "Timeout waiting for worker to respond" errors. Forks provide better isolation and stability in CI, while threads remain faster for local development.

- CI: forks pool (stable, no worker timeouts)
- Local: threads pool (faster)

Author: jdalton

.config/vitest.config.mts

@@ -70,8 +70,8 @@ const vitestConfig = defineConfig({
     ],
     reporters: ['default'],
     // Optimize test execution for speed
-    // Threads are faster than forks
-    pool: 'threads',
+    // Use forks in CI for stability, threads locally for speed
+    pool: process.env.CI ? 'forks' : 'threads',
     poolOptions: {
       threads: {
         // Maximize parallelism for speed
@@ -95,6 +95,13 @@ const vitestConfig = defineConfig({
         isolate: false,
         useAtomics: true,
       },
+      forks: {
+        // CI: Use forks for stability (no worker timeout issues)
+        singleFork: isCoverageEnabled,
+        maxForks: isCoverageEnabled ? 1 : 16,
+        minForks: isCoverageEnabled ? 1 : 4,
+        isolate: true,
+      },
     },
     // Increase timeouts to prevent worker timeout on slow CI environments
     teardownTimeout: 30_000,