Make the list of resolvers user-configurable

[Nour-Mws]

From #256 (comment):

Only provide options to enable/disable resolvers, not reorder them.

• Does not make sense to have ID mapping anywhere other than at the bottom of the stack
• Does not make sense to reorder PyPI vs --pyenv
• A user-provided mapping should always have the highest priority.

This can be done e.g via:

• Add a new option: --use-mappings that takes a sequence of "magic" words that correspond to the boxes in the diagram:
  • custom: [tool.fawltydeps.custom_mapping] from config file (if it exists). [^1]
  • pyenv: Any --pyenv option(s) given on the command or in the config file.
  • basepath: Any Python environments discovered under basepath i.e. in the current "project' (for lack of a better word). [^2]
  • current: The Python environment that FD is currently running in (aka sys.path).
  • pypi: The various components that make up the PyPI logic.
  • id: The identity mapping.
• The user can pass --use-mappingswith their own list to reorder and skip any of the mappings corresponding to the above words. In other words, advanced users have full control.
• The default value of --use-mappings (based on the above diagram) is: --use-mappings=custom,pyenv,basepath,current,id (pypi is off by default).
• Add a --use-pypi option to insert pypi before the id mapping in the default list, i.e. equivalent to --use-mappings=custom,pyenv,basepath,current,pypi,id
• Add a --use-pyenv-only option to reduce the list to --use-mappings=pyenv
• Consider adding more options to satisfy common user scenarios.
• Make all of these options mutually exclusive to prevent confusing scenarios with option combinations that don't make sense. (e.g. --use-pyenv-only --use-pypi).

[jherland]

Leaving this issue as not planned for now. We might want to revisit it later if/when we need more customizability in this area.