docs: add benchmarking framework and unified API reference (#415)

- Automated Benchmarking Framework (1,032 lines)
  - VSA, HSLM, FPGA benchmark categories with targets/thresholds
  - BenchmarkResult struct with regression detection
  - Multi-format output (JSON, Markdown, CSV)
  - CI/CD integration (GitHub Actions workflow)

- Unified API Reference (562 lines)
  - Part I: HSLM API (model, blocks, constants, attention)
  - Part II: VSA API (FHRR details, similarity metrics, usage examples)
  - Part III: FPGA API (synthesis, export formats, resource utilization)
  - Part IV: TRI-27 VM (registers, instruction set, memory model)
  - Part V: Research APIs (B2T, benchmarks, mining, training, CLI)
  - Part VI: CLI API (core commands, entry points)
  - Part VII: Queen API (bridge, perlexity, research agent)
  - Part VIII: Type System (standard, HybridBigInt, linear types, arena allocators)
  - Appendix A: Module dependencies
  - Appendix B: Quick reference cards (constants, VSA ops)
  - Usage examples: HSLM inference, VSA reasoning

- GitHub Actions workflow: benchmark CI, regression check, baseline management

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: Antigravity Agent