Agent Skills: benchmarking

Run and manage performance benchmarks with cargo xtask bench for facet-json, analyzing results with Markdown reports and comparing against serde_json baseline