FIX eagerLoad crash with nonterminal hasOne relation

[MasonD]

Addresses #11470

[MasonD]

Looks like the failing tests are because the number of queries is no longer as previously expected. I'd appreciate input on how what I've done here has caused extra queries.

[GuySartorelli]

Looks like the failing tests are because the number of queries is no longer as previously expected. I'd appreciate input on how what I've done here has caused extra queries.

My guess, having looked briefly at the code changes but not at the tests, is that the tests were actually not correct, and chains with has_one weren't actually checking anything after the has_one portion.
I don't have time to dive into the tests right now to validate that - can you please try to check whether the new results are correct or not?

[MasonD]

After investigation, the tests indeed did their job and caught a bug in my code, now fixed.

[MasonD]

@GuySartorelli I did resolve the failing tests, in case you missed it. The tests were correct and I was missing out on some eager fetching because of a bug.

[GuySartorelli]

Thanks yup. I saw that just haven't had a chance to come back to this yet.

[GuySartorelli]

Looks good and seems to work as expected, thanks for taking the time to do this.

Couple of things:

1. Can you please add some unit tests? One to check that this works as expected for HasOne.HasMany (i.e. the intended use case) and one to check that the exception is thrown for PolymorphicHasOne.HasMany?
2. Can you please retarget this to the 5.3 branch? That way the patch can be released immediately rather than waiting for the April minor release.
   You may need to reset your commits after retargeting the PR.

[MasonD]

@GuySartorelli all done!