Default_value option in Arguments inside Mutation is not working as expected

Start with the repeated schema.execute calls in the provided test_mutation reproduction and trace how the AnimalFarm Arguments default_value is passed into mutate. Confirm that an omitted animals argument receives a fresh empty list on each request, and add a regression test covering both executions.

Hi,

• What is the current behavior?
  When I am working with the following set up of graphene Mutation:

class AnimalFarm(graphene.Mutation):
    class Arguments:
        animals = graphene.List(graphene.String, default_value=[])

    output = graphene.List(graphene.String)

    def mutate(root, info, **filters):
        import pdb; pdb.set_trace()
        animals_mio = filters["animals"]
        animals_mio.append('duck')
        
        return animals_mio


class Mutation(graphene.ObjectType):
    animal_farm = AnimalFarm.Field()

where you can see that I am declaring a Mutation that have and argument called "animals" which type is "graphene.List(graphene.String, default_value=[])", that means if I dont pass this argument in the mutation, graphene automatically filled this argument ("animals") with an empty list.

At this point all works perfect, but when I try the following steps, I reckon that could be a little bug in to reset the old value of "animals" argument.

• Step 1: Install the latest version of graphene, in a linux OS, in my case is Ubuntu 22.04.3 LTS:

pip install "graphene>=3.1"

• Step 2: Copy this code in a file with extension ".py", for example:

nano cool_file_name.py

• Step 3: Copy this code to have the example, where you can this bug:

import graphene



class Query(graphene.ObjectType):
    nothing = graphene.String()
    
    def resolve_nothing(root, info):
        return "empty"

class AnimalFarm(graphene.Mutation):
    class Arguments:
        animals = graphene.List(graphene.String, default_value=[])

    output = graphene.List(graphene.String)

    def mutate(root, info, **filters):
        import pdb; pdb.set_trace()
        animals_local = filters["animals"]
        animals_local.append('duck')
        
        return animals_local


class Mutation(graphene.ObjectType):
    animal_farm = AnimalFarm.Field()


schema = graphene.Schema(query=Query, mutation=Mutation)

mutation = """
    mutation addAnimals{
        animalFarm {
            output
        }
    }
"""

def test_mutation():
    # First request
    result = schema.execute(mutation)
    print(result)

    # Second request, in this request it is the problem !!!
    result = schema.execute(mutation)
    print(result)

test_mutation()

• Step 4: Then save the changes and launch the file:

python3 cool_file_name.py

• Step 5: Execute "n" + Enter, to check this steps that works as expected in the first request, where we can see that both variables:
  animals_local
  filters["animals"]
  are equal as an empty list, as you can see in this image:
  
  Now if we add a value to the variable "aniamls_local" we also see that works perfect:
  

• Step 6: The bug appear in the second request, where we do same steps that in the Step 5, but in this second request if we check the value of varieble filters["animals"] we can see that its default value is the old value of the previous request, where we added "duck" word to "animals" list argument:
  

• What is the expected behavior?
  The expected behavior is that if I launch multiple request of this mutation and I dont pass any value for "animals" argument, the default_value must be what I declare in set up type (an empty list).
  So graphene have to reset correctly the default_value and not persist old values.

• Please tell us about your environment:

  • Version: graphene>=3.1
  • Platform: Ubuntu 22.04.3 LTS