Add performance comparison tests for dynamic vs generic operations

[konard]

Summary

This PR implements performance comparison tests and benchmarks to measure the performance difference between dynamic operations and other generic approaches, as requested in issue #68.

Implementation Details

Added Files

• ComparisonTests.cs: Unit tests comparing different generic addition approaches
• EqualityTests.cs: Unit tests comparing different generic equality approaches
• DynamicVsGenericBenchmarks.cs: Performance benchmarks using BenchmarkDotNet

Performance Approaches Compared

Based on the StackOverflow solution referenced in the issue (https://stackoverflow.com/a/8122675/710069):

1. INumber approach (NET 7+ - best performance)

   • Uses the new .NET 7+ generic math interfaces
   • Provides compile-time type safety and best runtime performance

2. EqualityComparer approach (good performance)

   • Uses EqualityComparer<T>.Default for equality comparisons
   • Well-optimized standard approach

3. Expression-based approach (moderate performance)

   • Compiles expressions at runtime and caches them
   • Good balance between flexibility and performance

4. Dynamic approach (slowest performance)

   • Uses C#'s dynamic keyword for operations
   • Most flexible but with significant performance overhead

Test Results

All tests pass successfully:

• ✅ ComparisonTests: Tests addition operations across all approaches
• ✅ EqualityTests: Tests equality operations across all approaches
• ✅ All existing tests continue to pass

Benchmark Usage

To run the performance benchmarks:

cd csharp/Platform.Data.Doublets.Benchmarks
dotnet run -c Release

The benchmarks will show the actual performance differences between:

• INumber addition vs Dynamic addition vs Expression addition
• IEqualityOperators equality vs Dynamic equality vs Expression equality vs EqualityComparer

Test plan

• Unit tests created and passing
• Benchmarks compile successfully
• All existing tests continue to pass
• Code follows project conventions

🤖 Generated with Claude Code

---

Resolves #68