Add performance benchmarks for type operations

[AliiiBenn]

Description

Add performance benchmarks for type manipulation operations.

Motivation

As the type manipulation system grows, we need to ensure it performs well, especially with:

• Large classes with many members
• Deeply nested type expressions
• Complex generic types

Tasks

• Create benchmark suite using pytest-benchmark or similar
• Benchmark key operations:
  • Iter[Attrs[T]] on classes with 10/50/100 members
  • GetMemberType calls
  • Type evaluation with nested generics
  • NewProtocol creation
• Document results and identify bottlenecks

Acceptance criteria

• Benchmark suite created
• Key operations benchmarked
• Results documented
• Performance issues identified (if any)