Add regression test for resultEqualityCheck receiving a cleared WeakRef

[veksa]

Closes #750

The bug reported in #750 was actually fixed a while back in cea19b9, where the old lastResult?.deref?.() ?? lastResult line was replaced with the maybeDeref helper. Once that shipped in 5.2.0 the runtime error is gone.

What was still missing is a test that locks the behavior in, so we don't accidentally regress it again. The original reporter tried to write one but couldn't get garbage collection to trigger reliably.

This adds that test. It builds a memoized selector, produces one result and registers it with a FinalizationRegistry, then drops every strong reference so the result becomes collectible. Instead of a single global.gc() call (which was flaky and just timed out), it runs a few GC passes and yields to the event loop between them until the finalization callback fires. At that point lastResult is a cleared WeakRef, and the test asserts that a subsequent call never hands that raw WeakRef to resultEqualityCheck.

The test is skipped when global.gc isn't exposed. I confirmed it fails against the old ?? lastResult logic and passes with the current maybeDeref implementation, so it genuinely guards the fix.

[netlify]

✅ Deploy Preview for reselect-docs canceled.

Name	Link
🔨 Latest commit	01f28f6
🔍 Latest deploy log	https://app.netlify.com/projects/reselect-docs/deploys/6a67065000d321000825a529

[codesandbox-ci]

This pull request is automatically built and testable in CodeSandbox.

To see build info of the built libraries, click here or the icon next to each commit SHA.