fix(auto-instrumentations-node): make SIGTERM shutdown test more reliable

[garysassano]

Which problem is this PR solving?

The test shuts down the NodeSDK when SIGTERM is received is flaky due to making HTTP requests to example.com:

Short description of the changes

The key improvements are:

1. Controlled Environment: Instead of depending on example.com (which could be slow/unreachable), we use a local server that responds immediately.

2. Better Event Handling: Changed from using req.on('close') to res.on('end') which is more reliable for confirming the request completed.

3. Proper Cleanup: Added server.close() to the SIGTERM handler to ensure clean shutdown.

4. Fixed Timing: The local server responds immediately, reducing the chance of hitting the test's 1500ms timeout.

The main issue was likely that requests to example.com were taking too long or failing, causing the test to hit its timeout before seeing the Finished request message. By using a local server, we make the test much more deterministic and faster.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 90.84%. Comparing base (3e95995) to head (ca846b1).
Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2667   +/-   ##
=======================================
  Coverage   90.84%   90.84%           
=======================================
  Files         172      172           
  Lines        8103     8103           
  Branches     1646     1646           
=======================================
  Hits         7361     7361           
  Misses        742      742           

[pichlermarc]

thanks, much appreciated 🙂