Floating point imprecision causes different dicts to be marked as similar if highly nested

Either we should migrate to integer only arithmetic OR use python decimal library to increase precision

I have a usecase where JSON schema are being marked similar if the difference is at deepest level in a fairly nested schema.

Alternatively instead of using similarity score < 1 to check for similarity we should maybe change criteria to be that returned diff is not empty?

                d, s = self._obj_diff(v, w)
                if s < 1.0:
                    changed[k] = d

i.e this code may need to change to instead check based on d instead of s

Provide feedback