Fix updating the GitHub Status as soon as any TC task fails

… rather than only when the entire task group is resolved.
This allows Homu to more quickly be notified of a failure,
and move on to the next PR in the queue sooner.

.taskcluster.yml

@@ -1,4 +1,8 @@
 version: 1
+
+# If and when switching to `reporting: checks-v1` here, also change the `statuses` route to `checks`
+# in `CONFIG.routes_for_all_subtasks` in `etc/taskcluster/decision_task.py`
+
 policy:
   # https://docs.taskcluster.net/docs/reference/integrations/taskcluster-github/docs/taskcluster-yml-v1#pull-requests
   pullRequests: public

etc/taskcluster/decision_task.py

@@ -21,6 +21,12 @@ def tasks(task_for):
             branch if not branch.startswith("try-") else "try"
         )
 
+    # Work around a tc-github bug/limitation:
+    # https://bugzilla.mozilla.org/show_bug.cgi?id=1548781#c4
+    if task_for.startswith("github"):
+        # https://github.com/taskcluster/taskcluster/blob/21f257dc8/services/github/config.yml#L14
+        CONFIG.routes_for_all_subtasks.append("statuses")
+
     # The magicleap build is broken until there's a surfman back end
     magicleap_dev = lambda: None
     magicleap_nightly = lambda: None