kill_retry_time does not work after `pm2 reload`

What's going wrong?

The kill_retry_time stop working after pm2 reload in cluster mode.

How could we reproduce this issue?

// ecosystem.config.js
{
      instances: '1',
      exec_mode: 'cluster',
      kill_timeout: 120000,
      kill_retry_time: 5000,
}

KILL without reload, work as expected

pm2 start
pm2 kill

============
  PM2                         | pid=2627 msg=failed to kill - retrying in 5000ms
  PM2                         | pid=2627 msg=failed to kill - retrying in 5000ms
============

KILL after reload, stop working

pm2 start

pm2 reload 1
============ LOGS
  PM2                         | pid=2627 msg=failed to kill - retrying in 100ms
  PM2                         | pid=2627 msg=failed to kill - retrying in 100ms
============


pm2 kill
============ LOGS
  PM2                         | pid=2627 msg=failed to kill - retrying in 100ms
  PM2                         | pid=2627 msg=failed to kill - retrying in 100ms
============

Supporting information

$ pm2 report

--- PM2 report ----------------------------------------------------------------
Date                 : Sat Feb 12 2022 18:29:57 GMT-0600 (Central Standard Time)
===============================================================================
--- Daemon -------------------------------------------------
pm2d version         : 5.1.2
node version         : 16.14.0
node path            : /var/folders/ft/8qnyv8812mz3bf1bylm7gqyr0000gn/T/fnm_multishells/71392_1644528424507/bin/pm2
argv0                : node
uid                  : 501
gid                  : 20
uptime               : 0min
===============================================================================
--- CLI ----------------------------------------------------
local pm2            : 5.1.2
node version         : 16.14.0
node path            : /var/folders/ft/8qnyv8812mz3bf1bylm7gqyr0000gn/T/fnm_multishells/71392_1644528424507/bin/pm2
argv0                : node
uid                  : 501
gid                  : 20
===============================================================================
--- System info --------------------------------------------
arch                 : arm64
platform             : darwin
type                 : Darwin
cpus                 : Apple M1 Pro
cpus nb              : 10
freemem              : 177553408
totalmem             : 34359738368
===============================================================================
--- PM2 list -----------------------------------------------
┌────┬────────────────────┬──────────┬──────┬───────────┬──────────┬──────────┐
│ id │ name               │ mode     │ ↺    │ status    │ cpu      │ memory   │
└────┴────────────────────┴──────────┴──────┴───────────┴──────────┴──────────┘
===============================================================================
PM2        | 2022-02-12T18:29:57: PM2 log: --- New PM2 Daemon started ----------------------------------------------------
PM2        | 2022-02-12T18:29:57: PM2 log: PM2 version          : 5.1.2
PM2        | 2022-02-12T18:29:57: PM2 log: Node.js version      : 16.14.0
PM2        | 2022-02-12T18:29:57: PM2 log: Current arch         : arm64
PM2        | 2022-02-12T18:29:57: PM2 log: PM2 home             : /Users//.pm2
PM2        | 2022-02-12T18:29:57: PM2 log: PM2 PID file         : /Users//.pm2/pm2.pid
PM2        | 2022-02-12T18:29:57: PM2 log: RPC socket file      : /Users//.pm2/rpc.sock
PM2        | 2022-02-12T18:29:57: PM2 log: BUS socket file      : /Users//.pm2/pub.sock
PM2        | 2022-02-12T18:29:57: PM2 log: Application log path : /Users//.pm2/logs
PM2        | 2022-02-12T18:29:57: PM2 log: Worker Interval      : 30000
PM2        | 2022-02-12T18:29:57: PM2 log: Process dump file    : /Users//.pm2/dump.pm2
PM2        | 2022-02-12T18:29:57: PM2 log: Concurrent actions   : 2
PM2        | 2022-02-12T18:29:57: PM2 log: SIGTERM timeout      : 1600
PM2        | 2022-02-12T18:29:57: PM2 log: ===============================================================================

Start by reproducing the issue with the ecosystem.config.js settings and the documented pm2 start, reload, and kill commands in cluster mode. Compare the retry interval before and after pm2 reload; done means kill_retry_time consistently produces 5000ms retries after reload as well as before it.