perf: parallel insert/hash performance and commit time regression

[AminR443]

Current State

#1258 introduced parallel insert and hash operations that improved propose times. However, commit times have regressed compared to the original implementation.

Goals

1. Reliable benchmarks to quantify the overall performance impact of parallel operations
2. Identify bottlenecks causing longer commit times
3. Determine optimal configuration and improvements to maximize performance gains

Tasks

• Create comprehensive benchmark suite comparing:
  • Original serial implementation
  • Current parallel implementation
  • Various parallelism configurations
• Profile commit phase to identify performance bottlenecks
• Document performance characteristics under different workloads
• Investigate potential optimizations for commit phase
• Provide configuration guidelines for best performance (perf: Use serial instead of parallel proposal creation if the batch size is < N #1368, chore: use bounded channel for parallel execution #1358, perf: If the number of CPUs is <8, disable parallel insertion code #1357)