DKAN-4287 Make harvest_run id not a primary key.

[swirtSJW]

Fixes #4287

Describe your changes

QA Steps

This list is INCOMPLETE at this time.

From existing data

• git checkout 2.18.3 (the tag before runs became an entity)
• init the site
• ddev drush dkan:sample-content:create
• ddev drush cron
  [] validate no errors show in terminal from these commands.

From new data

[] validate that the Run table list displays ___

• Add manual QA steps in checklist format for a reviewer to perform. Be as specific as possible, provide examples if appropriate.

Checklist before requesting review

If any of these are left unchecked, please provide an explanation

• I have updated or added tests to cover my code
• I have updated or added documentation

[swirtSJW]

It is possible that this can't be deprecated. Everywhere it is still called, there is no access to the actual ID. Mainly in tests where it is pulling previous specific run instances.

[swirtSJW]

There are still two test failures that are not immediately clear to me why they are failing

[swirtSJW]

Got it down to just one test failure
testGetExtractedUuids
Failed asserting that two arrays are equal.

[dafeder]

@swirtSJW I don't see any test failures at the moment?

[swirtSJW]

@paul-m Can you take a first pass at code review on this?
Tests are passing but I will hold off on adding QA steps until I perform this whole upgrade on PDC. Right now this is successful on vanilla but has not been fully run on an existing site.

[paul-m]

Under the 2.x branch, I used sample_content to generate a harvest to run (drush dkan:sample-content:create), and then ran it (using cron).
Then I switched to this branch and ran drush updb.
Everything seemed to work... I did notice that the one harvest run in my harvest_runs table didn't have a UUID.

So I went ahead and did a new installation with this branch, and then ran the sample content harvest, and it did have a UUID.

So perhaps the update process doesn't generate UUIDs the way just creating an entity does...

Also I'd suggest that the utility functions, such as harvest_get_temp_run_ids() in harvest.install should be in HarvestUtility. That way it's all in one place and labeled as part of the update process.

[swirtSJW]

Everything seemed to work... I did notice that the one harvest run in my harvest_runs table didn't have a UUID

So it looks like a simple database write did not kick off the magic that would add a uuid that a normal drupal save would. So I intentionally added it.

[swirtSJW]

I am stumped by this test failure.

There was 1 error:

1) Drupal\Tests\datastore\Unit\Plugin\QueueWorker\ImportJobTest::testBasics
TypeError: Drupal\datastore\Plugin\QueueWorker\ImportJob::getParser(): Return value must be of type Contracts\ParserInterface, CsvParser\Parser\Csv returned

It seems completely unrelated to the code in this PR.

Beyond this test @paul-m I made all the changes you requested.

[paul-m]

Over in #4177 we deprecated some interfaces, which led to the fails.