Improving Cursor Tab with online RL

Implemented online reinforcement learning to improve the Cursor Tab model. This involved defining a reward function that encourages accepted suggestions and penalizes rejected ones, and using the Policy Gradient Theorem to update the model based on user interactions. The system now deploys new model checkpoints frequently (1.5-2 hours) to collect on-policy data for continuous training, resulting in a new default Tab model that makes 21% fewer suggestions with a 28% higher accept rate.