test: evaluate resetForTests function

[jaissica12]

Background

• This PR is an effort to reduce flaky tests due to timing variance, and session timeouts.

What Has Changed

• Updated resetForTests function for:
  • Test-only defaults (isTestEnv): eventBatchingIntervalMillis, astBackgroundEvents, offlineStorage (and cleared window.onbeforeunload).
  • Safe teardown: cleared per-instance session globalTimer and global window.mParticle.forwardingStatsTimer.

Screenshots/Video

• {Include any screenshots or video demonstrating the new feature or fix, if applicable}

Checklist

• I have performed a self-review of my own code.
• I have made corresponding changes to the documentation.
• I have added tests that prove my fix is effective or that my feature works.
• I have tested this locally.

Additional Notes

• To keep scope of this PR limited this just updates resetForTests function.
• A follow up ticket SDKE-303 to this will update each test to clear timers/intervals.

Reference Issue (For employees only. Ignore if you are an outside contributor)

• Closes https://rokt.atlassian.net/browse/SDKE-302

[rmi22186]

Let's move this additional resetForTests functionality into the config/setup in the test folder to avoid increasing the size of the SDK for test purposes.

[rmi22186]

since I realize we use ._resetForTests all over the place, I'd expect it to be something like:

const originalRFT = mParticle._resetForTests;

mParticle._resetForTests = function() {
  // all this code here
  originalRFT();
}

This way all the future instances of ._resetForTests will run the code + the original _resetForTests function

[jaissica12]

Added additional code in setup.ts to keep everything together. I believe we're already resetting flags there and calling _resetForTests, so the additional code can be added here. Let me know thoughts

[rmi22186]

any reason this is null but line 230 is 0? I think we should set both to 0

[rmi22186]

what is the reason for setting this to null? seems like we should be setting it/mocking it for specific tests that require onbeforeunload.

[jaissica12]

Yes, individual tests like tests-beaconUpload.ts also set it to null before dispatching events, but having it in the global setup acts as a safeguard that ensures clean test isolation for both current and future tests.
When I started working on this I was exploring potential causes of flakiness, so this is just defensive approach even if a someone forgets to clear onbeforeunload in a specific test.

[rmi22186]

Hm, I think we should have it only in the test file it's being used. There's a chance an integration test depends on onbeforeunload in the future, and I'd rather be explicit about when it should be null and when it shouldn't. This can probably be part of a cleanup task.

[sonarqubecloud]

Quality Gate failed

Failed conditions
9.2% Duplication on New Code (required ≤ 3%)

See analysis details on SonarQube Cloud