fix tests on pypy

[nixjdm]

Currently there are two failing tests on pypy, shown below.

I didn't notice them because I also apparently have tox misconfigured, so I don't think I actually was using pypy in my local tests. In my branch to test github actions I have pypy running, where I noticed the failure. https://github.com/terminal-labs/deep_collections/actions/runs/3914234772 That's commented out for now. I'd like to add that back in when these tests are fixed.

2 pypy test failures:

____________________________ TestTuple.test_getitem ____________________________

self = <tests.test_builtins.TestTuple object at 0x0000000003bd5590>
dc = DeepCollection(('nested', ('thing', 'spam')))

    def test_getitem(self, dc):
        assert dc[0] == "nested"
        assert dc[1, 1] == "spam"
    
>       assert isinstance(dc[1], self._type)

tests/conftest.py:89: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
deep_collections/__init__.py:720: in __getitem__
    return DeepCollection(
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

cls = <class 'deep_collections.DeepCollection'>, args = ()
kwargs = {'match_args': (), 'match_kwargs': {}, 'match_with': 'glob', 'recursive_match_all': True}
obj = ('thing', 'spam'), dynamic_parent_cls = <class 'tuple'>
new_cls = <class 'deep_collections.DeepCollection_tuple'>

    def __call__(cls, *args, **kwargs):
        if args:
            obj = args[0]
            args = args[1:]
        else:
            obj = {}
    
        # If a DC is given, we must use the original type as the parent_cls. We don't
        # want or need to nest DC types, and we if we don't flattten the inheritence
        # we get a metaclass conflict can't produce a consistent method resolution.
        if hasattr(obj, "_obj") and isinstance(obj, DeepCollection):
            dynamic_parent_cls = type(obj._obj)
        else:
            dynamic_parent_cls = type(obj)
    
        # Make a new_cls that inherits from the dynamic_parent_cls, and resolve any
        # potential metaclass conflicts, like when the object has its own metaclass
        # already, as when it's an ABC.
        try:
            # Resultant type is deep_collections.DeepCollection_[type]
            new_cls = type(
                f"{cls.__name__}_{dynamic_parent_cls.__name__}",
                (cls, dynamic_parent_cls),
                {},
            )
        except TypeError:
            # Resolve metaclass conflict.
            # Create a new metaclass on the fly, merging the two we have.
            mcls = type(cls)
            cls.__class__ = type(
                f"{mcls.__name__}_{dynamic_parent_cls.__name__}",
                (mcls, type(dynamic_parent_cls)),
                {},
            )
            # Resultant type is likely abc.DeepCollection_[type]
            new_cls = type(
                f"{cls.__name__}_{dynamic_parent_cls.__name__}",
                (cls, dynamic_parent_cls),
                {},
            )
    
        # Create the instance and initialize it with the given object.
        # Sets obj in instance for immutables like tuple
        # instance = new_cls.__new__(new_cls, obj, *args, **kwargs)
>       instance = new_cls.__new__(new_cls, obj, *args, **kwargs)
E       TypeError: __new__() got 4 unexpected keyword arguments

deep_collections/__init__.py:557: TypeError
_____________________________ TestTuple.test_slice _____________________________

self = <tests.test_builtins.TestTuple object at 0x0000000003bd5050>

    def test_slice(self):
        dc = DeepCollection(self._type([*range(10)]))
>       assert dc[2:8] == self._type([2, 3, 4, 5, 6, 7])

tests/conftest.py:109: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
deep_collections/__init__.py:720: in __getitem__
    return DeepCollection(
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

cls = <class 'deep_collections.DeepCollection'>, args = ()
kwargs = {'match_args': (), 'match_kwargs': {}, 'match_with': 'glob', 'recursive_match_all': True}
obj = (2, 3, 4, 5, 6, 7), dynamic_parent_cls = <class 'tuple'>
new_cls = <class 'deep_collections.DeepCollection_tuple'>

    def __call__(cls, *args, **kwargs):
        if args:
            obj = args[0]
            args = args[1:]
        else:
            obj = {}
    
        # If a DC is given, we must use the original type as the parent_cls. We don't
        # want or need to nest DC types, and we if we don't flattten the inheritence
        # we get a metaclass conflict can't produce a consistent method resolution.
        if hasattr(obj, "_obj") and isinstance(obj, DeepCollection):
            dynamic_parent_cls = type(obj._obj)
        else:
            dynamic_parent_cls = type(obj)
    
        # Make a new_cls that inherits from the dynamic_parent_cls, and resolve any
        # potential metaclass conflicts, like when the object has its own metaclass
        # already, as when it's an ABC.
        try:
            # Resultant type is deep_collections.DeepCollection_[type]
            new_cls = type(
                f"{cls.__name__}_{dynamic_parent_cls.__name__}",
                (cls, dynamic_parent_cls),
                {},
            )
        except TypeError:
            # Resolve metaclass conflict.
            # Create a new metaclass on the fly, merging the two we have.
            mcls = type(cls)
            cls.__class__ = type(
                f"{mcls.__name__}_{dynamic_parent_cls.__name__}",
                (mcls, type(dynamic_parent_cls)),
                {},
            )
            # Resultant type is likely abc.DeepCollection_[type]
            new_cls = type(
                f"{cls.__name__}_{dynamic_parent_cls.__name__}",
                (cls, dynamic_parent_cls),
                {},
            )
    
        # Create the instance and initialize it with the given object.
        # Sets obj in instance for immutables like tuple
        # instance = new_cls.__new__(new_cls, obj, *args, **kwargs)
>       instance = new_cls.__new__(new_cls, obj, *args, **kwargs)
E       TypeError: __new__() got 4 unexpected keyword arguments