Use perform_with_capacity to schedule ReverificationBatchWorker

Geo::ReverificationBatchWorker uses the LimitedCapacity::Worker
concern. For that concern to work correctly, we must periodically
reschedule jobs and perform cleanup by calling perform_with_capacity.
Before this commit, we were never calling perform_with_capacity on
Geo::ReverificationBatchWorker, meaning we never ran the cleanup code.
This commit fixes that by replacing perform_async with
perform_with_capacity.

ee/app/models/concerns/geo/verifiable_replicator.rb

@@ -46,7 +46,7 @@ module Geo
 
         # Secondaries don't need to run this since they will receive an event for each
         # rechecksummed resource: https://gitlab.com/gitlab-org/gitlab/-/issues/13842
-        ::Geo::ReverificationBatchWorker.perform_async(replicable_name) if ::Gitlab::Geo.primary?
+        ::Geo::ReverificationBatchWorker.perform_with_capacity(replicable_name) if ::Gitlab::Geo.primary?
       end
 
       # Called by VerificationBatchWorker.

ee/spec/support/shared_examples/models/concerns/verifiable_replicator_shared_examples.rb

@@ -148,7 +148,7 @@ RSpec.shared_examples 'a verifiable replicator' do
         it 'does not enqueue ReverificationBatchWorker' do
           stub_secondary_node
 
-          expect(::Geo::ReverificationBatchWorker).not_to receive(:perform_async)
+          expect(::Geo::ReverificationBatchWorker).not_to receive(:perform_with_capacity)
 
           described_class.trigger_background_verification
         end
@@ -158,7 +158,7 @@ RSpec.shared_examples 'a verifiable replicator' do
         it 'enqueues ReverificationBatchWorker' do
           stub_primary_node
 
-          expect(::Geo::ReverificationBatchWorker).to receive(:perform_async).with(described_class.replicable_name)
+          expect(::Geo::ReverificationBatchWorker).to receive(:perform_with_capacity).with(described_class.replicable_name)
 
           described_class.trigger_background_verification
         end