chore: add beads ticket templates and bd-groom/bd-ripple scripts

chore: add .claude agents, commands, and consolidate CLAUDE.md

bd init: initialize beads issue tracking

chore: exclude benchmark artifacts from linting

docs: add context_size to benchmark methodology table

chore: sync beads issues

feat(benchmarks): run v2 provider comparison with pytest validation

chore: clean up v1 benchmark artifacts

feat(benchmarks): add LLMClient, MetricsCollector, and PytestValidator

feat: add Qwen3.5 benchmark suite and rename benchmark to perf

chore: sync beads

docs: fix score interpretation table in provider comparison

fix: update tests to match blank mode with no services

docs: clarify KLD ranking labels in provider comparison

chore: sync beads