Set the old primary host read_only=0 if failover/switch over failed.

[beigithub]

Hello:
For graceful switchover, we noticed in some cases, even if the candidate host is lagging, the orchestrator still goes ahead and set read_only=1 on the old primary before lag is noticed. This left the old primary with read_only =1 but switchover didn't happen, human intervention is needed to set read_only=0 on the old primary again. Is there a way to detect the lag and exit the switch-over process "before" setting read_only=1 on the old primary?