This pull request updates the way Python environments for Cellpose 3 and 4 detectors are defined and installed, moving from hardcoded YAML strings (and using micromamba) to TOML resource files. It also generalizes the environment installation logic to allow for different builders (e.g., Pixi or Mamba).

Environment definition and installation improvements:

• The environment definitions for Cellpose3 and Cellpose4 are now loaded from TOML resource files (cellpose3.toml, cellpose4.toml), allowing for platform-specific dependencies. Placeholders for environment name and Appose version are dynamically replaced at runtime.
• The environment installation utility (ApposeUtils.installEnvironment) is refactored to accept a Builder<?> parameter, making it possible to use either Pixi or Mamba for environment setup. All relevant calls and tests are updated to use this new signature.

UI and builder selection updates:

• The PythonEnvironmentManagerUI and its EnvironmentPanel now require and use a specific builder for each environment, enabling the use of Pixi for Cellpose and StarDist, and Mamba for Trackastra. This ensures the correct tool is used for each environment type.
• The Cellpose3 and Cellpose4 detectors now use Pixi for environment creation instead of Mamba, reflecting the move to TOML-based environments.

StarDist environment and config adjustments:

• The StarDist environment and configuration now use the environment name (ENV_NAME) instead of a hardcoded string, both for the model path and for extracting configuration from the Bioimage.io model descriptor.