feat: enable extended thinking for Claude models (anthropic/*)

fix: show partial-credit answers as yellow ~ in progress bar, not red incorrect

fix: leaderboard handles all-exam/all-year runs via summary.jsonl fallback

chore: remove unused io import in test_llm_interface.py

fix: encode question images as PNG (lossless) instead of JPEG

fix: failure_unexpected_type scores 0 not -1

Fix low-priority issues: dead code, parsing, deps, tests