Default keyword arguments result in error

[lilyminium]

In generate_default_metadata, the default parameter_gradient_keys=None, results in an error.

openff-evaluator/openff/evaluator/workflow/workflow.py

Lines 621 to 627 in 64d7440

	@staticmethod
	def generate_default_metadata(
	physical_property,
	force_field_path,
	parameter_gradient_keys=None,
	target_uncertainty=None,
	):

~/anaconda3/envs/polymetrizer/lib/python3.9/site-packages/openff/evaluator/workflow/workflow.py in generate_default_metadata(physical_property, force_field_path, parameter_gradient_keys, target_uncertainty)
    684         # Find only those gradient keys which will actually be relevant to the
    685         # property of interest
--> 686         relevant_gradient_keys = Workflow._find_relevant_gradient_keys(
    687             physical_property.substance, force_field_path, parameter_gradient_keys
    688         )

~/anaconda3/envs/polymetrizer/lib/python3.9/site-packages/openff/evaluator/workflow/workflow.py in _find_relevant_gradient_keys(substance, force_field_path, parameter_gradient_keys)
    568 
    569         # noinspection PyTypeChecker
--> 570         if parameter_gradient_keys == UNDEFINED or len(parameter_gradient_keys) == 0:
    571             return []
    572 

TypeError: object of type 'NoneType' has no len()

Ideally default values for functions should be ones that "just work". I'd suggest passing in parameter_gradient_keys=UNDEFINED.

[SimonBoothroyd]

Thanks for catching this - a PR with the fix would be welcome!