Refactor analytics.gather

With the change to use pk-based interval slicing for the job events
table, we need analytics.gather to be the code that manages all of the
"expensive" collector slicing.  While we are at it, let's ship each
chunked tarball file as we produce it.

awx/main/conf.py

@@ -10,6 +10,7 @@ from rest_framework.fields import FloatField
 
 # Tower
 from awx.conf import fields, register, register_validate
+from awx.main.fields import JSONField
 from awx.main.models import ExecutionEnvironment
 
 
@@ -778,6 +779,13 @@ register(
     category=_('System'),
     category_slug='system',
 )
+register(
+    'AUTOMATION_ANALYTICS_LAST_ENTRIES',
+    field_class=JSONField,
+    # label=_('Last gathered entries for expensive Automation Analytics collectors.'),
+    category=_('System'),
+    category_slug='system',
+)
 
 
 register(