Is your feature request related to a problem? Please describe.
The complexity of the maintenance windows, focusing on Recover strategy, can lead to a crash alert. Some user inputs, like the CEL, can cause a divergence and, therefore, a loss of alert.

As principal contributor to this strategy, I consider that the maintenance window flow shouldn't brake the main alert flow. Therefore, if the maintenance window makes a crash, the alert has to continue its principal flow.

This new behavior increases the alert stability.

Describe the solution you'd like
The solution is very simple, wrapping the check function in a try/except block, with the properly log.