Actuator should not delete Machine objects about cluster-api-provider-baremetal HOT 1 CLOSED

The code to delete the Machine was added in #51, so we will have to think carefully about how the use case addressed by that (which doesn't appear to be well documented) will still be addressed.

In the case where we're forcing the MachineSet to use all of the available Host inventory, going to the Failed phase should work fine: if a Host is deleted then the MachineSet will be scaled down and it is the unhealthy Machine that will be removed.

In the case where we're using the cluster autoscaler, things should work themselves out.

In the case where a MachineHealthCheck is operating, any Failed machines will be deleted (once the PR above is merged).

Only in the case where:

• There is excess Host inventory
• No autoscaling is enabled
• No MachineHealthCheck is enabled

would there be a change in behaviour. (Currently a new Machine would be created automatically, using one of the spare Hosts; if we don't delete the existing Machine then it will continue to exist in a Failed state until somebody deletes it manually, and no replacement would be created unless somebody manually deletes it or scales up the MachineSet. This would be consistent with how it works on other machine api providers.)

from cluster-api-provider-baremetal.