Adjust alert timing and severity

Currently, the alert timing is too aggressive and
severity is too high.  An individual machine being
down is not critical, any critical components already
have alerting configured (such as etcd).

install/0000_90_machine-api-operator_04_alertrules.yaml

@@ -15,19 +15,19 @@ spec:
         - alert: MachineWithoutValidNode
           expr: |
              (mapi_machine_created_timestamp_seconds unless on(node) kube_node_info) > 0
-          for: 10m
+          for: 60m
           labels:
-            severity: critical
+            severity: warning
           annotations:
             message: "machine {{ $labels.name }} does not have valid node reference"
     - name: machine-with-no-running-phase
       rules:
         - alert: MachineWithNoRunningPhase
           expr: |
             (mapi_machine_created_timestamp_seconds{phase!="Running"}) > 0
-          for: 10m
+          for: 60m
           labels:
-            severity: critical
+            severity: warning
           annotations:
             message: "machine {{ $labels.name }} is in phase: {{ $labels.phase }}"
     - name: machine-api-operator-metrics-collector-up