evaluator rules for ha workflow

When large scale failure in cloud platform, if triggering all recovery workflow at once, it maybe lead to further damage to the cloud platform. So the users usually expects to stop HA, to avoid further damage to the cloud platform.

There is no direct detection for large scale failure, but it could use methods of data aggregated analysis. It needs to introduces an evaluator and a list of rules, which works likes a fuse protector. It will prevent chaotically recovery workflow, when larger scale failure.