Making new Python repls 100x faster to start up

We've improved the Python experience on Replit by addressing issues such as slow package installation times and large package sizes. A new Python template has been released, which uses nix for faster package installation, a standard virtual environment, and caching to reduce disk space usage. The caching mechanism stores files in a content-addressable cache, allowing for fast installation of packages even with large dependencies like TensorFlow. This update also aims to improve the overall developer experience by reducing start-up times for Python repls, and plans to extend this treatment to other programming languages as well.