Add performance timing and comprehensive documentation

[oulrich1]

Performance Logging:

• Add timing to test_training.cpp
• Track training time separately from testing time
• Display millisecond precision timing for both phases
• Helps identify performance bottlenecks

Documentation Updates (README.md):

• Add "Running Toy Example Training" section
• Document all 4 training tests (Linear Regression, XOR, AND, OR)
• Include architecture details and training parameters
• Add build optimization instructions (Debug vs Release)
• Document expected performance differences (10-60x speedup)
• Include PyTorch comparison instructions
• Provide timing output examples

This enables users to:

1. Easily run toy examples locally
2. Measure and debug performance
3. Compare with optimized builds
4. Benchmark against PyTorch baseline