Improve signal propagation on kubernetes executor when `FF_USE_DUMB_INIT_WITH_KUBERNETES_EXECUTOR` is enabled

Summary

In the MR !4443 (merged) and !4485 (merged), we added support for dumb-init in the job container. However, the propagation of signals still needs improvement.

This issue is a follow-up of the thread started here: #27443 (comment 1652493554) to investigate and improve the current behavior.

Steps to reproduce

.gitlab-ci.yml

variables:
  FF_PRINT_POD_EVENTS: "true"
  FF_USE_POWERSHELL_PATH_RESOLVER: "true"
  FF_SCRIPT_SECTIONS: "true"
  FF_USE_DUMB_INIT_WITH_KUBERNETES_EXECUTOR: "true"
  CI_DEBUG_SERVICES: "true"

job:
  image: alpine:3.18
  script: |
    pstree -p
    ps -ef
    sleep 3600

Actual behavior

The Pod is not terminated

Expected behavior

The Pod should be terminated

Relevant logs and/or screenshots

NA

Environment description

config.toml contents

concurrent = 1
check_interval = 1
log_level = "debug"
shutdown_timeout = 0

listen_address = ':9252'

[session_server]
  session_timeout = 1800

[[runners]]
  name = ""
  url = "https://gitlab.com/"
  id = 0
  token = "glrt-REDACTED"
  token_obtained_at = "0001-01-01T00:00:00Z"
  token_expires_at = "0001-01-01T00:00:00Z"
  executor = "kubernetes"
  shell = "bash"
  [runners.kubernetes]
    host = ""
    bearer_token_overwrite_allowed = false
    image = "alpine"
    pull_policy = "always"
    allowed_pull_policies = ["always", "if-not-present"]
    pod_termination_grace_period_seconds = 3600
    namespace = ""
    namespace_overwrite_allowed = ""
    pod_labels_overwrite_allowed = ""
    service_account_overwrite_allowed = ""
    pod_annotations_overwrite_allowed = ""
    node_selector_overwrite_allowed = ".*"
    allow_privilege_escalation = false
    [[runners.kubernetes.services]]
    [runners.kubernetes.dns_config]
    [runners.kubernetes.pod_labels]
      user = "ratchade"

Used GitLab Runner version

Possible fixes