Q learning algorithm

[anudeep-gad12]

• Replace UCB logic with a Q-learning engine for adaptive question flow
• Implement state-action-reward loop using Combos and InteractionLogs tables
• Insert real curriculum questions with difficulty, choices, and correct_answer
• Wire question selection to Q-learning output and update Q-values per answer
• Log emotions, hints, speed, and correctness to shape reward signal
• Display next AI-recommended action and update UI based on it