cleanup: fix some typos in code comment

lib/fluent/log.rb

@@ -606,7 +606,7 @@ def terminate
     end
   end
 
-  # This class delegetes some methods which are used in `Fluent::Logger` to a instance variable(`dev`) in `Logger::LogDevice` class
+  # This class delegates some methods which are used in `Fluent::Logger` to a instance variable(`dev`) in `Logger::LogDevice` class
   # https://github.com/ruby/ruby/blob/7b2d47132ff8ee950b0f978ab772dee868d9f1b0/lib/logger.rb#L661
   class LogDeviceIO < ::Logger::LogDevice
     def flush

lib/fluent/root_agent.rb

@@ -201,7 +201,7 @@ def lifecycle(desc: false, kind_callback: nil)
     def start
       lifecycle(desc: true) do |i| # instance
         i.start unless i.started?
-        # Input#start sometimes emits lots of evetns with in_tail/`read_from_head true` case
+        # Input#start sometimes emits lots of events with in_tail/`read_from_head true` case
         # and it causes deadlock for small buffer/queue output. To avoid such problem,
         # buffer related output threads should be run before `Input#start`.
         # This is why after_start should be called immediately after start call.

test/command/test_fluentd.rb

@@ -807,7 +807,7 @@ def multi_workers_ready?
       )
     end
 
-    test 'success to start workers when configured plugins as a chidren of MultiOutput only for specific worker do not support multi worker configuration' do
+    test 'success to start workers when configured plugins as a children of MultiOutput only for specific worker do not support multi worker configuration' do
       script = <<-EOC
 require 'fluent/plugin/output'
 module Fluent::Plugin

test/plugin/test_in_tail.rb

@@ -1011,7 +1011,7 @@ def test_unwatched_files_should_be_removed
       waiting(5) { sleep 0.1 until d.instance.instance_variable_get(:@tails).keys.size == 0 }
 
       # Previous implementation has an infinite watcher creation bug.
-      # Following code checks such unexpected bug by couting  actual object allocation.
+      # Following code checks such unexpected bug by counting actual object allocation.
       base_num = count_timer_object
       2.times {
         sleep 1