Fix `COUNT(DISTINCT ...)` for `GROUP BY` with `ORDER BY` and `LIMIT`

This is the fix for the regression of #29848.

In #29848, I've kept existing select list in the subquery for the count
if ORDER BY is given. But it had accidentally affect to GROUP BY
queries also. It should keep the previous behavior in that case.

Fixes #30886.

activerecord/CHANGELOG.md

+*   Fix `COUNT(DISTINCT ...)` for `GROUP BY` with `ORDER BY` and `LIMIT`.
+
+    Fixes #30886.
+
+    *Ryuta Kamizono*
+
+
 ## Rails 5.1.4 (September 07, 2017) ##
 
 *   No changes.

activerecord/lib/active_record/relation/calculations.rb

@@ -195,7 +195,7 @@ def perform_calculation(operation, column_name)
         if operation == "count"
           column_name ||= select_for_count
           if column_name == :all
-            if distinct && !(has_limit_or_offset? && order_values.any?)
+            if distinct && (group_values.any? || !(has_limit_or_offset? && order_values.any?))
               column_name = primary_key
             end
           elsif column_name =~ /\s*DISTINCT[\s(]+/i

activerecord/test/cases/calculations_test.rb

@@ -265,6 +265,10 @@ def test_distinct_joins_count_with_order_and_limit_and_offset
     assert_equal 3, Account.joins(:firm).distinct.order(:firm_id).limit(3).offset(2).count
   end
 
+  def test_distinct_count_with_group_by_and_order_and_limit
+    assert_equal({ 6 => 2 }, Account.group(:firm_id).distinct.order("1 DESC").limit(1).count)
+  end
+
   def test_should_group_by_summed_field_having_condition
     c = Account.group(:firm_id).having("sum(credit_limit) > 50").sum(:credit_limit)
     assert_nil        c[1]