Refactor utility functions for performance and maintainability

[rjrodger]

Summary

This PR refactors several utility functions in src/util.ts to improve performance, code clarity, and maintainability. The changes focus on optimizing recursive operations, replacing array lookups with Set-based checks, and improving the circular reference detection algorithm.

Key Changes

• Refactored dive() function: Extracted the recursive logic into a new diveInternal() helper function that uses an explicit prefix array parameter instead of manipulating array indices. This eliminates the need for unshift() operations and reduces array allocations.

• Added hasOwnKeys() utility: Introduced a new helper function to check if an object has enumerable properties, replacing the Object.keys(child).length === 0 check for better performance.

• Optimized order_sort() function: Converted key_order array lookups to use a Set for O(1) lookup performance instead of O(n) array includes checks.

• Optimized order_exclude() and order_include() functions: Replaced Array.includes() calls with Set.has() for better performance when filtering items.

• Improved decircular() function: Refactored to use a shared path array that is mutated during traversal (with push/pop) instead of creating new arrays at each recursion level. This reduces memory allocations and improves performance for deeply nested objects.

• Minor fix in entity() function: Changed Object.entries().map() to Object.entries().forEach() since the return value was not being used.

• Added comprehensive tests: Added test cases for stringify() and decircular() functions to ensure correctness of the refactored code.

Implementation Details

The diveInternal() refactoring maintains the same behavior while improving efficiency by:

• Passing the accumulated path as a parameter instead of reconstructing it
• Using a for-of loop instead of reduce for better readability
• Avoiding array mutations on the result paths

The decircular() optimization uses a closure-scoped path array that is mutated during traversal, significantly reducing garbage collection pressure for large object graphs.

https://claude.ai/code/session_017xSbcASVkxy5qHKKzn7kKi