Update Model.__eq__ logic

[macisamuele]

Model equality check was implemented by ensuring that two objects of the same class and with the same "dict" representation are considered the same.

As far as this definition is valid we might do it a bit better.
Currently, isinstance and issubclass checks do consider equivalent model types generated by different bravado_core.spec.Spec instances if they are created out of the same specs (same url and content).
So what can happen is that

instance1 = spec1.definitions['model1'](...)
instance2 = spec2.definitions['model1']model1(...)

assert isinstance(instance1, instance2.__class__)
assert isinstance(instance2, instance1.__class__)
assert instance1._as_dict() == instance2._as_dict()
assert instance1 == instance2  # This would fail with the currently released bravado-core and succeed after this PR

The added test, test_model_equality_if_model_class_generated_by_different_Spec_object, will ensure that regression will be catched if the behaviour will change and it currently fails if the changes in bravado_core/model.py are reverted

[coveralls]

Coverage decreased (-0.1%) to 98.895% when pulling 820ed6c on macisamuele:maci-update-equality-condition-of-models into d92cab7 on Yelp:master.

[sjaensch]

I don't see how this comment relates to the check you're doing, or the issue you're fixing with this PR. How does this help with two different Model classes (which are for the same model)?

[macisamuele]

I was initially implementing the check as if not isinstance(other, self.__class__): return False but I was noticing that in case of polymorphic models (definitions that take advantage of polymorphism) we are not short-circuiting the evaluation at the "type" level.

But the more I read it, the more I think that we could be fine on not short-circuiting the evaluation as different types would have different value of the discriminator field and so the subsequent check would catch such cases.

Said so, I'm going to simplify this so we remove eventual ambiguity.