fix: revert idle socket validation to setTimeout(0) to prevent stall on idle event loop

[marceli1404]

Summary

Reverts scheduleIdleSocketValidation in client-h1.js from setImmediate() back to setTimeout(0), along with the corresponding clearImmediate → clearTimeout.

Background

PR #5499 changed setTimeout(0) → setImmediate() to remove the ~1ms timer floor on idle socket validation. While that change performed well under sustained load (where the event loop is continuously busy), it causes up to ~500ms stalls when reusing an idle keep-alive socket on an otherwise idle event loop.

Root Cause

When a keep-alive socket is reused and the event loop is otherwise idle:

1. The previous request completes → resumeH1() unrefs the socket (socket.unref())
2. The next request is queued → scheduleIdleSocketValidation sets an unref()'d setImmediate
3. The socket is re-ref'd but the setImmediate fires in the check phase (after poll)
4. Since the socket was just ref'd but has no pending I/O, the poll phase waits for its default timeout before proceeding to the check phase — this can be up to 500ms on some platforms

setTimeout(0) fires in the timers phase (before poll), so the callback runs promptly regardless of poll timeout, at the cost of ~1ms timer resolution.

Reproduction

Script in the issue shows sequential requests on a 1-connection pool taking 20ms, 458ms, 34ms, 51ms, 450ms, 464ms instead of ~13ms each.

Fix

This PR reverts the approach from #5499. The original motivation (removing the timer floor) is worthwhile, but it introduces an order-of-magnitude worse regression on idle event loops. A hybrid approach (check-phase scheduling with poll-timeout prevention) could be explored separately.

Fixes #5600

[mcollina]

Can you add a test?

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.49%. Comparing base (8453a77) to head (b1ac19a).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #5606   +/-   ##
=======================================
  Coverage   93.49%   93.49%           
=======================================
  Files         110      110           
  Lines       38281    38303   +22     
=======================================
+ Hits        35789    35812   +23     
+ Misses       2492     2491    -1     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[marceli1404]

Added a regression test for the idle socket reuse stall ( est/node-test/keep-alive-reuse.js). It makes 6 sequential requests on a single-connection pool with 50ms server delay and asserts each request completes within 200ms. The test passes on this branch.

[mcollina]

LGTM. The setTimeout(0) revert is the right fix for the idle keep-alive stall, and the regression test now drives progress via server connection/request events instead of wall-clock thresholds.