Unify extraction of alert message in Prometheus probe

[bzurkowski]

Currently, only message is expected. Other message fields cause probe or ingestor failure. We should support other common message fields e.g. summary:

{
   "status":"firing",
   "labels":{
      "alertname":"PrometheusRuleFailures",
      "endpoint":"web",
      "instance":"10.233.96.145:9090",
      "job":"prometheus-prometheus-oper-prometheus",
      "namespace":"monitoring",
      "pod":"prometheus-prometheus-prometheus-oper-prometheus-0",
      "prometheus":"monitoring/prometheus-prometheus-oper-prometheus",
      "service":"prometheus-prometheus-oper-prometheus",
      "severity":"critical"
   },
   "annotations":{
      "description":"Prometheus monitoring/prometheus-prometheus-prometheus-oper-prometheus-0 has failed to evaluate 60 rules in the last 5m.",
      "summary":"Prometheus is failing rule evaluations."
   },
   "startsAt":"2020-04-27T11:11:22.598Z",
   "endsAt":"0001-01-01T00:00:00Z",
   "generatorURL":"http://prometheus-prometheus-oper-prometheus.monitoring:9090/graph?g0.expr=increase%28prometheus_rule_evaluation_failures_total%7Bjob%3D%22prometheus-prometheus-oper-prometheus%22%2Cnamespace%3D%22monitoring%22%7D%5B5m%5D%29+%3E+0&g0.tab=1",
   "fingerprint":"515b3d41b2a970fe"
}