fix: enhance resolution and enum handling in anyOf schemas

[Edison-A-N]

Summary

Fix $ref resolution and enum handling in anyOf schemas. Resolves issue #897 where enum types defined via $ref in anyOf were not properly resolved.

Motivation and Context

Addresses issue #897: anyOf=[$ref, null] patterns were not resolved, causing enum properties to be lost and preventing dropdown rendering.

How Has This Been Tested?

• All existing tests pass (392 unit tests, 16 E2E tests)
• Prettier and ESLint checks passed
• Tested with schemas containing anyOf=[{$ref: "#/$defs/UserRole"}, {type: "null"}]

Breaking Changes

None. Bug fix maintaining backward compatibility.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation update

Checklist

• I have read the MCP Documentation
• My code follows the repository's style guidelines
• New and existing tests pass locally
• I have added appropriate error handling
• I have added or updated documentation as needed

Additional context

• Enhanced resolveRef: Added circular reference detection, recursively resolves $ref in anyOf
• Optimized normalizeUnionType: Unified handling for anyOf=[type, null], preserves enum properties
• Integrated in ToolsTab: Uses resolveRef before normalization to ensure complete schema info

Fixes #897

[olaservo]

Hi @Edison-A-N thanks for your PR! There's a spicy take about supporting refs in tool schemas here: #755

So think that this would apply to elicitation schemas as well. I'll drop a comment there since I hadn't responded there yet. Please feel free to share your thoughts as well.

[cliffhall]

Hi @Edison-A-N!

This looks like it builds upon #889 from last week, I see you are calling the resolveRef function it added. I also see it replaces some changes from #895 which I also made last week for nullable arrays, rolling up all the nullable type checks into a new block, which will be nice if it didn't break anything. I need to test that out and make sure nullable items still work correctly first.