fog: Try ipmi power-cycle if stuck in a reimage reboot hang

[djgalloway]

Fixes: https://tracker.ceph.com/issues/74717

Works: https://pulpito.ceph.com/dgalloway-2026-02-12_21:01:11-rados:cephadm-wip-rocky10-branch-of-the-day-2026-02-03-1770151121-distro-default-trial/

2026-02-12T21:03:43.012 INFO:teuthology.provision.fog.trial163:Waiting for deploy to finish
2026-02-12T21:03:43.047 INFO:teuthology.provision.fog.trial020:Waiting for deploy to finish

2026-02-12T21:06:53.165 INFO:teuthology.provision.fog.trial020:Deploy complete!

...

2026-02-12T21:12:28.627 DEBUG:teuthology.orchestra.connection:{'hostname': 'trial163.front.sepia.ceph.com', 'username': 'ubuntu', 'timeout': 60}
2026-02-12T21:12:31.709 WARNING:teuthology.provision.fog.trial163:[Errno None] Unable to connect to port 22 on 10.20.193.163
2026-02-12T21:12:37.715 WARNING:teuthology.provision.fog.trial163:trial163: SSH not up after 427s (~70% of timeout); power cycling and continuing to wait
2026-02-12T21:12:37.716 INFO:teuthology.orchestra.console.trial163:Power off
2026-02-12T21:12:37.716 DEBUG:teuthology.orchestra.console.trial163:pexpect command: ipmitool -H trial163.ipmi.sepia.ceph.com -I lanplus -U inktank -P ApGNXcA7 power off
2026-02-12T21:12:37.758 DEBUG:teuthology.orchestra.console.trial163:power off output: Chassis Power Control: Down/Off
2026-02-12T21:12:41.762 DEBUG:teuthology.orchestra.console.trial163:pexpect command: ipmitool -H trial163.ipmi.sepia.ceph.com -I lanplus -U inktank -P ApGNXcA7 power status
2026-02-12T21:12:41.873 DEBUG:teuthology.orchestra.console.trial163:check power output: Chassis Power is off
2026-02-12T21:12:41.873 INFO:teuthology.orchestra.console.trial163:Power off completed
2026-02-12T21:12:41.973 INFO:teuthology.orchestra.console.trial163:Power on
2026-02-12T21:12:41.974 DEBUG:teuthology.orchestra.console.trial163:pexpect command: ipmitool -H trial163.ipmi.sepia.ceph.com -I lanplus -U inktank -P ApGNXcA7 power on
2026-02-12T21:12:42.014 DEBUG:teuthology.orchestra.console.trial163:power on output: Chassis Power Control: Up/On
2026-02-12T21:12:46.019 DEBUG:teuthology.orchestra.console.trial163:pexpect command: ipmitool -H trial163.ipmi.sepia.ceph.com -I lanplus -U inktank -P ApGNXcA7 power status
2026-02-12T21:12:46.130 DEBUG:teuthology.orchestra.console.trial163:check power output: Chassis Power is on
2026-02-12T21:12:46.130 INFO:teuthology.orchestra.console.trial163:Power on completed
2026-02-12T21:12:46.231 DEBUG:teuthology.orchestra.connection:{'hostname': 'trial163.front.sepia.ceph.com', 'username': 'ubuntu', 'timeout': 60}
2026-02-12T21:12:49.309 WARNING:teuthology.provision.fog.trial163:[Errno None] Unable to connect to port 22 on 10.20.193.163

...

2026-02-12T21:14:49.526 INFO:teuthology.provision.fog.trial163:Deploy complete!

I used 70% in the test job but that's way too long so changed to 50%. I simulated a reboot hang by powering trial163 off after I saw it start booting after reimage.

[dmick]

I got the PR title. Please change the commit msg.

[dmick]

Don't see any reason we shouldn't include this, even if other reboot methods prove more reliable