feat: add Qwen3.5 model catalog and agentic evaluation framework

Models:
- configs/models.conf: catalog with Qwen3.5-35B-A3B (MoE, top pick),
  Qwen3.5-27B (dense), Qwen3-Coder-30B-A3B (agentic/coding)
- Updated benchmark setup to show catalog with download status
- docs/model-recommendations.md: memory planning, quantization guide

Agentic evaluation:
- scripts/agentic/setup.sh: installs inspect-ai, evalplus, bigcodebench
  in a Python venv
- scripts/agentic/run-eval.sh: runs evaluations against local LLM server
  (ollama or llama.cpp). Suites: quick (HumanEval+IFEval), code
  (EvalPlus+BigCodeBench), tooluse (BFCL), full (all)
- bin/agentic: dispatcher with help
- docs/agentic-benchmarks.md: methodology, framework comparison, model
  recommendations for agentic use

Updated: Makefile (6 new targets), README, CLAUDE.md, docs/references.md

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

bin/agentic

@@ -0,0 +1,30 @@
+#!/usr/bin/env bash
+# Agentic evaluation dispatcher
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
+
+case "${1:-help}" in
+    setup)   exec bash "$SCRIPT_DIR/scripts/agentic/setup.sh" ;;
+    run)     exec bash "$SCRIPT_DIR/scripts/agentic/run-eval.sh" "${@:2}" ;;
+    quick)   exec bash "$SCRIPT_DIR/scripts/agentic/run-eval.sh" --suite quick "${@:2}" ;;
+    code)    exec bash "$SCRIPT_DIR/scripts/agentic/run-eval.sh" --suite code "${@:2}" ;;
+    tooluse) exec bash "$SCRIPT_DIR/scripts/agentic/run-eval.sh" --suite tooluse "${@:2}" ;;
+    full)    exec bash "$SCRIPT_DIR/scripts/agentic/run-eval.sh" --suite full "${@:2}" ;;
+    *)
+        echo "Usage: agentic <command> [options]"
+        echo ""
+        echo "Commands:"
+        echo "  setup     Install evaluation frameworks (inspect-ai, evalplus, bigcodebench)"
+        echo "  quick     EvalPlus HumanEval+ + IFEval (~1 hour)"
+        echo "  code      EvalPlus + BigCodeBench (~2-3 hours)"
+        echo "  tooluse   BFCL function calling evaluation (~1-2 hours)"
+        echo "  full      All evaluations (~5-6 hours)"
+        echo "  run       Custom run (--suite SUITE --model NAME --endpoint URL)"
+        echo ""
+        echo "All commands require --model NAME. Examples:"
+        echo "  agentic quick --model qwen3.5:35b-a3b-q8_0"
+        echo "  agentic full --model qwen3-coder:30b-a3b --endpoint http://localhost:8080/v1"
+        exit 1
+        ;;
+esac