Address review comments by tigerwnz

Refactors the test to share the namespace and feature flag
context

Improves the batch calculation to be more readable than using
fdiv

ee/app/workers/clear_shared_runners_minutes_worker.rb

@@ -20,10 +20,7 @@ class ClearSharedRunnersMinutesWorker # rubocop:disable Scalability/IdempotentWo
       start_id = Namespace.minimum(:id)
       last_id = Namespace.maximum(:id)
 
-      # Uses float division to avoid a ZeroDivisionError
-      # when namespace ID range is lower than BATCH_SIZE
-      # See: https://gitlab.com/gitlab-org/gitlab/-/issues/330068
-      batches = (last_id - start_id).fdiv(BATCH_SIZE)
+      batches = [(last_id - start_id) / BATCH_SIZE, 1].max
       execution_offset = (TIME_SPREAD / batches).to_i
 
       (start_id..last_id).step(BATCH_SIZE).with_index do |batch_start_id, batch_index|

ee/spec/workers/clear_shared_runners_minutes_worker_spec.rb

@@ -133,7 +133,7 @@ RSpec.describe ClearSharedRunnersMinutesWorker do
       end
     end
 
-    context 'when ci_parallel_minutes_reset feature flag is enabled, with batch size lower than namespace count' do
+    context 'when ci_parallel_minutes_reset feature flag is enabled' do
       subject { worker.perform }
 
       before do
@@ -142,40 +142,33 @@ RSpec.describe ClearSharedRunnersMinutesWorker do
         [2, 3, 4, 5, 7, 8, 10, 14].each do |id|
           create(:namespace, id: id)
         end
-
-        # Test with a batch size lower than count of namespaces
-        stub_const("#{described_class}::BATCH_SIZE", 3)
       end
 
-      it 'runs a worker per batch', :aggregate_failures do
-        # Spread evenly accross 8 hours (28,800 seconds)
-        expect(Ci::BatchResetMinutesWorker).to receive(:perform_in).with(0.seconds, 2, 4)
-        expect(Ci::BatchResetMinutesWorker).to receive(:perform_in).with(7200.seconds, 5, 7)
-        expect(Ci::BatchResetMinutesWorker).to receive(:perform_in).with(14400.seconds, 8, 10)
-        expect(Ci::BatchResetMinutesWorker).to receive(:perform_in).with(21600.seconds, 11, 13)
-        expect(Ci::BatchResetMinutesWorker).to receive(:perform_in).with(28800.seconds, 14, 16)
-
-        subject
-      end
-    end
+      context 'with batch size lower than count of namespaces' do
+        before do
+          stub_const("#{described_class}::BATCH_SIZE", 3)
+        end
 
-    context 'when ci_parallel_minutes_reset feature flag is enabled, with batch size higher than namespace count' do
-      subject { worker.perform }
+        it 'runs a worker per batch', :aggregate_failures do
+          # Spreads evenly accross 8 hours (28,800 seconds)
+          expect(Ci::BatchResetMinutesWorker).to receive(:perform_in).with(0.seconds, 2, 4)
+          expect(Ci::BatchResetMinutesWorker).to receive(:perform_in).with(7200.seconds, 5, 7)
+          expect(Ci::BatchResetMinutesWorker).to receive(:perform_in).with(14400.seconds, 8, 10)
+          expect(Ci::BatchResetMinutesWorker).to receive(:perform_in).with(21600.seconds, 11, 13)
+          expect(Ci::BatchResetMinutesWorker).to receive(:perform_in).with(28800.seconds, 14, 16)
 
-      before do
-        stub_feature_flags(ci_parallel_minutes_reset: true)
-        [2, 3, 4, 5, 7, 8, 10, 14].each do |id|
-          create(:namespace, id: id)
+          subject
         end
-
-        # Use the default BATCH_SIZE (100_000)
       end
 
-      it 'runs the worker in a single batch', :aggregate_failures do
-        # Runs a single batch, immediately
-        expect(Ci::BatchResetMinutesWorker).to receive(:perform_in).with(0.seconds, 2, 100_001)
+      context 'with batch size higher than count of namespaces' do
+        # Uses default BATCH_SIZE
+        it 'runs the worker in a single batch', :aggregate_failures do
+          # Runs a single batch, immediately
+          expect(Ci::BatchResetMinutesWorker).to receive(:perform_in).with(0.seconds, 2, 100001)
 
-        subject
+          subject
+        end
       end
     end
   end