Implement System.Runtime.Intrinsics for BitString operations

[konard]

Summary

This PR implements System.Runtime.Intrinsics support for BitString operations as requested in issue #59, providing hardware-accelerated SIMD operations for better performance.

Changes

Core Implementation

• Added intrinsics-based methods: IntrinsicsNot(), IntrinsicsAnd(), IntrinsicsOr(), IntrinsicsXor() and their parallel variants
• AVX2 and SSE2 support: Automatically detects and uses AVX2 when available, falls back to SSE2, then to regular operations
• Unsafe optimizations: Uses unsafe fixed pointers for direct memory access with SIMD instructions
• Smart fallbacks: Gracefully handles systems without intrinsics support

Performance Features

• AVX2 operations: Process 4 long values (256 bits) simultaneously when supported
• SSE2 operations: Process 2 long values (128 bits) simultaneously as fallback
• Parallel variants: Combine intrinsics with parallel processing for maximum throughput
• Optimized loops: Vectorized processing with scalar cleanup for remainder elements

Testing & Validation

• Comprehensive test coverage: All intrinsics methods tested for correctness against regular operations
• Benchmark integration: Added benchmark methods to measure performance improvements
• Experiment validation: Created test script that verifies intrinsics produce identical results

Technical Details

• Hardware detection: Runtime checks for Avx2.IsSupported and Sse2.IsSupported
• Memory safety: Proper fixed pointer usage with bounds checking
• Border optimization: Maintains existing border tracking optimizations
• Thread safety: Parallel variants use proper partitioning

Test Results

All existing tests pass plus new intrinsics-specific tests. The experiment script confirms:

• ✅ All operations produce identical results to regular implementations
• ✅ Proper fallback behavior when AVX2/SSE2 not available
• ✅ Correct handling of different BitString sizes

Performance Impact

The intrinsics implementations should provide significant performance improvements:

• AVX2: Up to 4x faster for large BitString operations
• SSE2: Up to 2x faster when AVX2 not available
• Parallel: Additional speedup on multi-core systems
• No regression: Falls back to existing implementations when needed

Usage

var bitString = new BitString(100000);
// Use intrinsics-accelerated operations
bitString.IntrinsicsNot();
bitString.IntrinsicsAnd(otherBitString);
bitString.ParallelIntrinsicsOr(otherBitString);

This implementation fulfills issue #59's request to explore System.Runtime.Intrinsics for BitString performance improvements.

🤖 Generated with Claude Code

---

Resolves #59