Code Review

This pull request updates Python build dependencies (pip and setuptools) and optimizes the model/dataset download process. Key improvements include implementing a retry mechanism for hash validation failures, introducing server-side prefix filtering and paginated filtering to reduce memory usage during snapshot downloads, and increasing the buffer size for hash computations. Additionally, the download process now includes fault-tolerant reporting, though feedback highlights the need to ensure that pagination errors and partial download failures trigger exceptions to prevent inconsistent snapshot states.