refactor(migrations): optimize icon migration performance

[kamja44]

Summary

Optimizes icon migration performance by replacing O(n²) array search with O(1) Map lookups.

Problem

The current implementation uses Object.values().find() for every imported icon, resulting in O(n²) time complexity. With large icon mappings (500+ icons), this becomes noticeably slow.

Solution

• Build a reverse index map (sourceToTargetMap) once at the start
• Use Map.get() for O(1) lookups instead of array searches
• Reduces complexity from O(n×m) to O(n+m)

Performance Impact

Example: 50 imported icons, 500 total mappings

• Before: 25,000 operations (O(n²))
• After: 550 operations (O(n))
• ~45x faster

Changes

• Added sourceToTargetMap construction before processing imports
• Replaced Object.values().find() with Map.get()
• Removed TODO comment about O(n²) performance

Testing

• Existing tests pass (no behavior changes)
• No new tests needed (internal optimization only)

Checklist

• Code follows project conventions
• Commit message follows conventional commits
• No breaking changes

Resolves TODO comment at line 164.

[vercel]

@kamja44 is attempting to deploy a commit to the shadcn-pro Team on Vercel.

A member of the Team first needs to authorize it.