Many Python versions, one Bazel build

In a complex build environment involving migrations between different Python runtime versions or the integration of Bazel, managing multiple Python interpreters within one build graph can be challenging. The process described involves creating a system that allows for the use of multiple Python interpreters by intercepting the default interpreter execution and substituting it with the required version, using a stub script generated from a template. This involves setting up a py_runtime definition, utilizing the expand_template rule from the aspect_bazel_lib library, and configuring a stub template to manage interpreter paths via environment variables. The strategy includes defining macros for py_binary and py_test to allow specifying different interpreter versions, which are stored in a structured format for easy reference. The implementation also considers the management of external dependencies, ensuring that the pip_install rule aligns with the correct interpreter version to avoid mismatched dependencies. This approach is particularly useful when there are no overlapping dependencies between targets requiring different Python versions, facilitating a seamless transition and flexible environment setup.