Promote maxUnhealthy: Investigate and potentially encourage users to adopt maxUnhealthy with fixed values (e.g., maxUnhealthy=5) as a more scalable alternative to minHealthy.

Implement stormRecoveryThreshold: Introduce a new parameter, tentatively named stormRecoveryThreshold. Once the number of healthy nodes drops below minHealthy, fencing should be delayed until the number of unhealthy nodes reaches this stormRecoveryThreshold. This would prevent premature fencing during transient issues and potential "storm" scenarios.