Always resetErrorInfo if structuredTypeRelatedTo succeeds

[jakebailey]

Fixes #49475

If typeRelatedToDiscriminatedType succeeds, then we have to clear the errors from previous steps as they are no longer true. Most of the previous steps do this too (but not all, confusingly).

This manifested as being "uncomment this code, and the error goes away" because the call expression's call to isRelatedTo had reportErrors=false, while the class' call had reportErrors=true. The class check is deferred, so the call expression happened first, and the successful relationship was cached, then the error didn't appear later. Without the call expression, the error was reported even though it shouldn't have been.

Rather than remembering to place those resetErrorInfo calls everywhere, turn the function into a worker and always reset if the relation succeeded.

[jakebailey]

@typescript-bot test this
@typescript-bot user test this inline

[typescript-bot]

Heya @jakebailey, I'm starting to run the diff-based user code test suite on this PR at b77d3e9. Hold tight - I'll update this comment with the log link once the build has been queued.

[typescript-bot]

Heya @jakebailey, I'm starting to run the extended test suite on this PR at b77d3e9. Hold tight - I'll update this comment with the log link once the build has been queued.

[RyanCavanaugh]

I guess my question is whether there are other "missing" calls in this function, e.g. at https://github.com/microsoft/TypeScript/pull/49718/files#diff-d9ab6589e714c71e657f601cf30ff51dfc607fc98419bf72e04f6b0fa92cc4b8L19836

[jakebailey]

I totally agree, and so I made the change to wrap structuredTypeRelatedTo and always restore on failure, but that seemed to break things in confusing ways. I'll go post that as a separate PR based on this one to show what I mean.

[typescript-bot]

Heya @jakebailey, I've run the RWC suite on this PR - assuming you're on the TS core team, you can view the resulting diff here.

[jakebailey]

All of the differences were my mistake; I'll push the change here. But, I don't have any more tests for cases I'm missing. At least it'll feel like it won't be broken :D

[jakebailey]

saveErrorInfo is kept because there are still some calls which explicitly reset it in order to hide certain errors that would be less helpful than others.