fix: pre-translate non-English chunks to English before LLM inference

fix: enable sampling — temperature=0.8 for MCQ, 0.4 for evaluator

fix: stronger MCQ/evaluator prompts — no duplicates, no self-reference, English-only, section 2 always present

fix: make the llama.cpp CPU runtime actually usable on the Space

fix: force English output in all LLM prompts regardless of PDF language

fix: prompts évaluateur séparés EN/FR pour forcer la langue du feedback

fix: texte question/feedback invisible + prompt langue renforcé

feat: switch to MiniCPM4-8B, add robustness, score tracker, language selector

feat: initial project structure with all core modules