Add performance benchmarks with criterion to buswatch-sdk

[lowhung]

Closes #17

Summary

This PR adds comprehensive performance benchmarks to buswatch-sdk using Criterion.rs to measure and track performance of critical SDK operations.

Changes

• ✅ Add criterion 0.5 as dev-dependency with html_reports feature
• ✅ Create 4 benchmark suites covering all requirements from test: Add performance benchmarks with criterion #17:
  • record_operations: Measures record_read() and record_write() hot path latency
  • snapshot_collection: Measures collect() snapshot generation time with varying module/topic counts
  • concurrent_throughput: Measures concurrent increment throughput across multiple threads
  • serialization: Measures JSON serialization and deserialization performance
• ✅ Add [[bench]] sections to Cargo.toml for each benchmark suite
• ✅ Update README.md with benchmark documentation and usage examples
• ✅ Update CHANGELOG.md with benchmark feature details

Benchmark Coverage

record_operations

• Single operation latency for record_read and record_write
• Varying message counts (1, 10, 100, 1000)
• Mixed read/write operations
• Multiple topics

snapshot_collection

• Varying topic counts (1, 5, 10, 50, 100)
• Varying module counts (1, 5, 10, 20, 50)
• Realistic workload scenarios
• Empty snapshots (baseline)

concurrent_throughput

• Varying thread counts (1, 2, 4, 8, 16)
• High contention (same topic) vs low contention (different topics)
• Mixed concurrent operations
• Concurrent collect() calls

serialization

• Varying snapshot sizes (small, medium, large, xlarge)
• Pretty-printed JSON
• Deserialization
• Round-trip serialization
• Realistic application scenarios

Usage

# Run all benchmarks
cargo bench

# Run specific benchmark suite
cargo bench --bench record_operations
cargo bench --bench snapshot_collection
cargo bench --bench concurrent_throughput
cargo bench --bench serialization

Results are saved to target/criterion/ with HTML reports for detailed analysis.

Testing

• ✅ All benchmarks compile successfully
• ✅ All benchmarks run without errors (tested with --test flag)
• ✅ All existing tests pass
• ✅ Documentation builds correctly