Populate sprint lookup incrementally on copy

Assigns `state.sprint_id_lookup` after each created sprint (not once
after the loop) so a mid-loop `create!` failure leaves the partial
mapping intact instead of nil, which would otherwise make copied work
packages retain the source project's sprint. Per-iteration assignment
is required because `state` is a `Hashie::Mash`: assigning the empty
hash before the loop converts it into a separate `Mash` instance, so
later mutations of the plain `Hash` are invisible to `state`. Also
strengthens the sprint-copy spec with a source/target decoupling
assertion.

Author: myabc

modules/backlogs/app/services/projects/copy/sprints_dependent_service.rb

@@ -47,9 +47,14 @@ def copy_dependency(*)
         attributes = source_sprint.attributes.dup.except("id", "project_id", "created_at", "updated_at")
         sprint = target.sprints.create!(attributes)
         sprint_id_map[source_sprint.id] = sprint.id
-      end
 
-      state.sprint_id_lookup = sprint_id_map
+        # Re-assigned on every iteration (rather than once after the loop) because `state` is a
+        # Hashie::Mash: assignment converts the Hash into a new Mash, breaking object identity, so
+        # later mutations of `sprint_id_map` alone would not be visible through `state.sprint_id_lookup`.
+        # Keeping the lookup current after each sprint also means a mid-loop `create!` failure (rescued
+        # by `Copy::Dependency#perform`) leaves the partial mapping intact instead of nil.
+        state.sprint_id_lookup = sprint_id_map
+      end
     end
   end
 end

modules/backlogs/spec/services/projects/copy_service_integration_spec.rb

@@ -144,6 +144,9 @@
       copied_sprint = project_copy.sprints.first
       expect(copied_sprint.id).not_to eq(source_sprint.id)
       expect(copied_sprint.project).to eq(project_copy)
+
+      copied_sprint.update!(name: "Renamed Copy")
+      expect(source_sprint.reload.name).to eq("Sprint A")
     end
   end