Refactor the import legacy persist logic

[jonathangoulding]

We have recently introduced functionality to import data from NALD using for a licence and companies for that particular licence.

This change will break out the persist functionality into logical groups of functionality.

[jonathangoulding]

I think it's worth having the big test to test the complete transaction.

Do we want to add duplicate tests per persist service ?

Worth a discussion.

[Cruikshanks]

I think it's worth having the big test to test the complete transaction.

Do we want to add duplicate tests per persist service ?

Worth a discussion.

Looks like we need a discussion! For me, the main driver for this was just how big and complex the tests were becoming, which is a 'smell' that the service under test is doing too much.

So, I would expect to see tests specifically for the things we have broken out, and persist-import.service.test.js to feature a lot of stubbing of those child services.

The child services can then focus on ensuring they behave as expected when inserting or updating. In PersistImportService we could, for example, start looking at how it behaves when one of those child services errors, i.e. confirm that nothing gets persisted. 🤷

[Cruikshanks]

Just gonna ping these back and then will check in again later, as they'll also help make the changes a bit less confusing for me!