What is the recommended method to install a Python library for all notebooks on a cluster?

Installing a Python library for all notebooks on a cluster in Databricks is efficiently and uniformly handled through the "Compute > Create > Libraries > Install New" method. This approach allows you to add libraries directly to the cluster, ensuring that they are available to all notebooks that are running on that specific cluster.

By using this method, you benefit from a centralized way to manage your libraries, which reduces the overhead of installing them individually for each notebook. It streamlines the workflow and helps maintain consistency across different notebooks since any updates or changes to the library can be managed at the cluster level.

The other options focus on either command-line methods, package management, or attaching libraries specifically to individual notebooks, which do not provide the same level of accessibility and convenience across the entire cluster as installing via the Libraries interface does. Therefore, choosing to install through the Libraries option ensures that every notebook can leverage the same library without the need for repeated installations or configurations.