Specify “most recent” aggregation mode for all gauge metrics (#16)

This change updates all of the gauges to include "most recent" aggregation mode introduced in prometheus-client 2.1.0. This is safe for both single- and multi-process mode because only the DirectFileStore from prometheus-client handles the aggregation option -- there is no aggregation to be done in single-process mode or with other adapters.

Fixes #15

lib/puma/plugin/yabeda.rb

@@ -15,15 +15,15 @@ def start(launcher)
     Yabeda.configure do
       group :puma
 
-      gauge :backlog, tags: %i[index], comment: 'Number of established but unaccepted connections in the backlog'
-      gauge :running, tags: %i[index], comment: 'Number of running worker threads'
-      gauge :pool_capacity, tags: %i[index], comment: 'Number of allocatable worker threads'
-      gauge :max_threads, tags: %i[index], comment: 'Maximum number of worker threads'
+      gauge :backlog, tags: %i[index], comment: 'Number of established but unaccepted connections in the backlog', aggregation: :most_recent
+      gauge :running, tags: %i[index], comment: 'Number of running worker threads', aggregation: :most_recent
+      gauge :pool_capacity, tags: %i[index], comment: 'Number of allocatable worker threads', aggregation: :most_recent
+      gauge :max_threads, tags: %i[index], comment: 'Maximum number of worker threads', aggregation: :most_recent
 
       if clustered
-        gauge :workers, comment: 'Number of configured workers'
-        gauge :booted_workers, comment: 'Number of booted workers'
-        gauge :old_workers, comment: 'Number of old workers'
+        gauge :workers, comment: 'Number of configured workers', aggregation: :most_recent
+        gauge :booted_workers, comment: 'Number of booted workers', aggregation: :most_recent
+        gauge :old_workers, comment: 'Number of old workers', aggregation: :most_recent
       end
 
       collect do