fix: isolate dind cleanup subprocesses from Ctrl+C SIGINT

[Copilot]

Pressing Ctrl+C broadcasts SIGINT to the entire foreground process group, which killed the docker stop/docker rm/docker network rm subprocesses in Stop() before they could complete — leaving the dind container and network alive and producing the cascade of errors.

Changes

• run() helper — calls setSysProcAttr(cmd), which sets Setpgid: true on Unix, placing each cleanup subprocess in its own process group and shielding it from the terminal's SIGINT
• Stop() — replaces two-step docker stop + docker rm with atomic docker rm -f, eliminating the race window between the two calls
• sysattr_unix.go / sysattr_windows.go — platform build tags to keep syscall.SysProcAttr out of the Windows build path

func run(name string, args ...string) (string, error) {
    cmd := exec.Command(name, args...)
    setSysProcAttr(cmd) // Setpgid: true on Unix — new process group, no SIGINT propagation
    out, err := cmd.CombinedOutput()
    return string(out), err
}

Original prompt

---

This section details on the original issue you should resolve

<issue_title>error when exiting with ctrl+c</issue_title>
<issue_description>^Cconstruct: stop dind: exit status 130

construct: rm dind: exit status 1
Error response from daemon: cannot remove container "construct-dind-8d8cdb6c7c7cdbf6": container is running: stop the container before removing or force remove

construct: rm network: exit status 1
Error response from daemon: error while removing network: network construct-net-8d8cdb6c7c7cdbf6 has active endpoints (name:"construct-dind-8d8cdb6c7c7cdbf6" id:"86facf98d546")
exit status 1
</issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes error when exiting with ctrl+c #2

---

🔒 GitHub Advanced Security automatically protects Copilot coding agent pull requests. You can protect all pull requests by enabling Advanced Security for your repositories. Learn more about Advanced Security.