What is the Best Approach to Dynamically Set a Default Environment to use with Pixi Tasks?

[JackEAllen]

When developing an application that must operate across multiple environments - such as with or without CUDA, depending on the availability of a CUDA-compatible GPU - it would be useful to have a straightforward method for toggling between which environment is treated as default. For instance, switching between default = ["runtime", "dev", "test", "lint"] and default = ["runtime", "dev", "gpu", "test", "lint"] would simplify the execution of tasks without explicitly needing to specify an environment (e.g. pixi run mytask instead of pixi run -e cpu-env mytask or pixi run -e gpu-env mytask).

What is the recommended approach to implement this functionality without modifying pixi.toml?
Additionally, is it possible to configure this through .pixi/config.toml so that individual environment preferences can be excluded from version control via .gitignore?