test(datastore): Add OptimisticConcurrency tests

[stocaaro]

Description of changes

Outcomes:

• Update graphql service fake to mimic OptimisticConcurrency(OCC) conflict resolution
• Add OptimisticConcurrency (OCC) tests for all cases covered by the existing Automerge tests
• Update service call cont expectation to reflect multiple event types, provide more rich feedback for failures

Description of how you validated changes

By comparing sample app results to test expectations and iterating until they align.

There is existing odd behavior where errors occur at in low latency conditions where the retry message includes null values for fields, which overwrites external updates. I will do sample app testing to confirm that this is consistent with the real world behavior.

Related changes

1. refactor(datastore): Move conflict resolution tests to their own file #12728
2. refactor(datastore): Refactor conflict resolution tests optimize for concise/readable #12732
3. refactor(datastore): Refactor test names and make subscription events more sample app like #12740
4. 🌱 test(datastore): Add OptimisticConcurrency tests
5. fix(datastore): Update outbox comparison logic to fix update versioning problems stocaaro/amplify-js#87

Checklist

• PR description included
• yarn test passes
• Tests are changed or added

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[david-mcafee]

Overall this looks great! I do think removing the incorrect comment is a blocker, since it would lead to significant confusion (at least for me!) down the line. Other than that, just one correction to a typo :)

[stocaaro]

@david-mcafee - Appreciate the suggestions. I've accepted both suggested changes and have re-read all of the comments to make sure nothing else thats coming across is auto-merge centric.

[david-mcafee]

LGTM! 🎉

[svidgen]

This else branch looks like a no-op. Since the fake is branching on canOccMerge() first anyway, I think I'd prefer that this throw to alert us that "we shouldn't be here", unless I'm failing to see or grok a use-case for this branch.

I think this is the only area of the PR that concerns me at the moment. Unless I'm just having a slow brain day, can we either turn this into a throw or add a comment explaining the branch?

[stocaaro]

Rereading this, I think this branch will never be reached since I have extracted the canOccMerge function, we're essentially doing the version check twice. I'll get a refactor out around this.