Do not notify the perf sheriff team anymore when the raptor decision task fails. Also be clearer about who should look into the failure first

.taskcluster.yml

@@ -240,7 +240,7 @@ tasks:
                         $if: 'trust_level == 3'
                         then:
                           - notify.email.fenix-eng-notifications@mozilla.com.on-failed
-                          - notify.email.perf-sheriffs@mozilla.com.on-failed
+                          - notify.email.perftest-alerts@mozilla.com.on-failed
                       payload:
                         command:
                           - >-
@@ -252,6 +252,13 @@ tasks:
                         cron: {$json: {$eval: 'cron'}}
                         treeherder:
                           symbol: raptor-D
+                        notify:
+                          email:
+                            link:
+                              text: Treeherder Job
+                              href: "https://treeherder.mozilla.org/#/jobs?repo=${project_name}&revision=${head_rev}"
+                            subject: "[${project_name}] Raptor decision task failed"
+                            content: "This calls for an action of the development team. If needed, escalate to the release engineering team in #releaseduty-mobile on Slack. Use the link to view it on Treeherder."
                       metadata:
                         name: ${project_name} Raptor Decision Task
                         description: Decision task scheduled by cron task [${cron.task_id}](https://tools.taskcluster.net/tasks/${cron.task_id})