Consider removing dict baseclass

[nickstanisha]

Because of NestedDict's behavior when indexing tuples, the following situation could arise for a user of this project

def my_user_function(dictobject):
    assert(isinstance(dictobject, dict)), "I'm a function that only works on dicts"
    # start indexing tuples into dictobject
    return dictobject

dictobject == nesteddictobject  # True
my_user_function(dictobject) == my_user_function(nesteddictobject)  # False

In response to this dilemma, I could either

• Do nothing -- relying on the user to distinguish use-cases using isinstance(obj, NestedDict)
• Use an internal self._dict object instead of subclassing dict

Using approach 2, the following conditions would have to be satisfied 1) isinstance(obj, dict) == False 2) Attributes not found in NestedDict will default to attributes in the internal self._dict (effectively replicating the current class inheritance)

• getattr(NestedDict, attr) should never retrieve the default attrs that NestedDict gets by virtue of being a Python object (e.g. __eq__, __contains__, etc.). The available attributes should only reflect what is hard-coded in nesteddict.py and what's present in dir(dict()) 3) import json; json.dumps(nesteddict) still works (nice to have) 4) dict(nesteddict) and NestedDict(dict) still work

[nickstanisha]

@meowklaski Feel free to develop and make pull requests against the no_inheritance branch. I've already added the minimum level of functionality to pass all of the tests, but I'm not satisfied with it yet because it doesn't fully satisfy point 2 above.

Once the no_inheritance branch is ready, I'll do a final assessment of whether or not the additional code complexity is worth avoiding the headache in the code sample above.