Correct docs, logic for model_construct behavior with extra

[sydney-runkle]

Fix #8266

1. Correct + expand on docs re model_construct behavior based on a model's extra setting
2. Correct behavior in the case of model_config.extra == 'ignore'

Selected Reviewer: @adriangb

[cloudflare-pages]

Deploying with    Cloudflare Pages

Latest commit:	b195e7d
Status:	✅  Deploy successful!
Preview URL:	https://ad0bef4f.pydantic-docs2.pages.dev
Branch Preview URL:	https://fix-model-construct-doc.pydantic-docs2.pages.dev

View logs

[codspeed-hq]

CodSpeed Performance Report

Merging #8807 will not alter performance

_{Comparing fix-model-construct-doc (b195e7d) with main (071c2c8)}

Summary

✅ 10 untouched benchmarks

[sydney-runkle]

Please review

[adriangb]

This makes sense, but like you say it is technically a breaking change. Is there a strong justification to make this change?

[sydney-runkle]

This makes sense, but like you say it is technically a breaking change. Is there a strong justification to make this change?

I think it's a minor enough breaking change that we can include it in a minor release, and I'd rather have strong documentation + correct behavior sooner rather than later. The sooner we fix this, the fewer people a change would impact down the line. But other than one person reporting the inconsistency in the docs / behavior, there's no strong justification.

[adriangb]

I think the fact that the runtime behavior is different than the documented behavior is a good start to justify the change.