Make setting and retrieving pydocstyle settings less tedious

[AlexWaygood]

Summary

This PR is stacked on top of #12581.

The pydocstyle rules have settings that allow you to specify a list of decorators that should be considered property-like, and a list of decorators that should be ignored. The list of property-like decorators isn't just read by the pydocstyle rules; it's also used by many other rules so that they can pass in "extra" property-like decorators to ruff_python_semantic::analyze::visibility::is_property().

Currently the pydocstyle Settings struct exposes these two settings as BTreeSet<String>, however, which is somewhat useless. To do any analysis with these values, rules first have to convert these lists into lists of QualifiedNames, which is tedious and verbose. This PR therefore makes the ignore_decorators and property_decorators fields on ruff_linter::rules::pydocstyle::settings::Settings private. Instead, it exposes public methods for setting the fields, and public methods for lazily iterating over the property-decorators or ignored-decoratos where the iterators map each decorator to a QualifiedName. The PR also adjusts ruff_python_semantic::analyze::visibility::is_property() so that it is able to receive a lazy iterator of QualifiedNames for the extra_property field, rather than a slice of QualifiedNames.

Test Plan

No new tests added since this is a pure refactor that shouldn't change semantics at all.

[github-actions]

ruff-ecosystem results

Linter (stable)

✅ ecosystem check detected no linter changes.

Linter (preview)

✅ ecosystem check detected no linter changes.