Agent Eval: Clean up legacy eval framework (9,200 lines of dead code)

[kovtcharov]

Problem

The legacy eval framework (eval.py at 159.7 KB, groundtruth.py, batch_experiment.py, transcript_generator.py, email_generator.py) still exists alongside the new agent eval framework. This is 9,200+ lines of dead code that:

• Confuses new contributors ("which eval framework do I use?")
• Inflates the codebase
• Has stale CLI commands (gaia eval, gaia groundtruth, gaia report) that conflict with the new gaia eval agent

What to Remove

Per #573 (the replacement plan):

Files to Remove

• src/gaia/eval/eval.py (3,336 lines) — old Evaluator class
• src/gaia/eval/groundtruth.py (~1,000 lines) — old ground truth generator
• src/gaia/eval/batch_experiment.py (2,367 lines) — old batch runner
• src/gaia/eval/transcript_generator.py — not needed
• src/gaia/eval/email_generator.py — not needed
• src/gaia/eval/fix_code_testbench/ — replaced by eval scenarios
• src/gaia/eval/configs/ — old config format
• src/gaia/eval/webapp/ — old Express.js visualizer (if superseded)

Files to Keep

• src/gaia/eval/runner.py — new AgentEvalRunner ✅
• src/gaia/eval/scorecard.py — new scorecard ✅
• src/gaia/eval/audit.py — new architecture audit ✅
• src/gaia/eval/claude.py — ClaudeClient (Anthropic SDK wrapper) ✅
• src/gaia/eval/config.py — MODEL_PRICING + DEFAULT_CLAUDE_MODEL ✅
• src/gaia/eval/pdf_document_generator.py → rename to pdf_generator.py ✅

CLI Changes

• Remove: gaia eval (old), gaia groundtruth, gaia report, gaia create-template, gaia visualize
• Keep: gaia eval agent (new framework)

Acceptance Criteria

• Legacy eval files removed
• Old CLI commands removed
• gaia eval agent remains the single entry point
• No broken imports after cleanup
• Tests updated to reflect removal