SocketPool+kQueue: retry kevent on EINTR instead of failing

[ianegordon]

Summary

Applies the same fix as a196c2e (ePoll) to the kQueue backend. kevent(2) documents: "[EINTR] A signal was delivered before the timeout expired and before any events were placed on the kqueue for return." No events are lost — but getNotifications() treated -1/EINTR as fatal, unwinding SocketPool.run() whose defer { cancelAll() } cancels every waiting continuation. A single signal delivery (e.g. Xcode's debugger pausing/resuming the process) tore down the whole pool on Darwin.

Now returns no events so the caller polls again, using the same guard shape and comment as the ePoll fix.

The Poll backend has the same gap, but a fix there interacts with Windows/WSAPoll semantics — see #228 for details and a proposed direction.

Testing

Full suite passes (449 tests, 51 suites). No new test: forcing EINTR requires delivering a signal to the exact thread blocked in kevent, which isn't deterministically arrangeable under Swift Concurrency (the ePoll fix in a196c2e shipped without a test for the same reason).

🤖 Generated with Claude Code

[codecov]

Codecov Report

❌ Patch coverage is 75.00000% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 92.89%. Comparing base (4e246d3) to head (58af82c).

Files with missing lines	Patch %	Lines
FlyingSocks/Sources/SocketPool+kQueue.swift	75.00%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #229   +/-   ##
=======================================
  Coverage   92.88%   92.89%           
=======================================
  Files          71       71           
  Lines        3727     3731    +4     
=======================================
+ Hits         3462     3466    +4     
  Misses        265      265           

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