🚀 Optimization: Reduce predict() complexity from O(n²) to O(n)

Fixes #74

Problem

The existing implementation iterates over each row and repeatedly filters the dataframe, resulting in O(n²) time complexity and poor scalability.

Solution

Replaced the nested iteration with pandas groupby(["True X", "True Y"]), ensuring each group is processed exactly once.

Improvements

• Reduced effective time complexity from repeated filtering (O(n²)) to grouped processing (O(n))
• Eliminated redundant dataframe filtering
• Improved readability and maintainability

Additional Fix

• Fixed indentation issue that caused incorrect dictionary population

Testing

• Successfully tested /api/session/batch_predict endpoint using Postman
• Verified correct JSON response
• No regressions observed