test(jest): recycle workers above 512MB to bound vm-context module registry growth (#3550)

Adds `workerIdleMemoryLimit: '512MB'` to jest.config.js. Jest forks worker
child processes that accumulate per-suite vm-context module registry under
`--experimental-vm-modules` (~14 MB retained per loaded suite). On large
downstream test suites this leaks until the worker OOMs.

The flag tells jest to inspect each worker's RSS between suites and respawn
when it exceeds the threshold. Respawn cost (~150-300 ms) is negligible vs.
the OOM crash it avoids on long-running coverage runs.

This complements PR #3549 (migration pre-warm) which addressed migration
overhead but not the vm-context module registry growth.

Author: PierreBrisorgueil

jest.config.js

@@ -221,4 +221,11 @@ export default {
 
   // Whether to use watchman for file crawling
   // watchman: true,
+
+  // Recycle worker child processes when their RSS exceeds this threshold.
+  // Counters jest's per-vm-context module registry accumulation (~14 MB
+  // retained per loaded suite under --experimental-vm-modules) which would
+  // otherwise OOM long-lived workers on large downstream test suites.
+  // The respawn cost (~150-300 ms) is negligible vs. the OOM crash it avoids.
+  workerIdleMemoryLimit: '512MB',
 };