Data is lost if unflatten column name equals to another original column

[manycoding]

Seems like a corner case, so I think at least we should raise an error when such situation happens:

echo '{"id": [0], "id_0": 2}' | flatten_json
>>>{"id_0": 2}

As you see, "id": [0] is lost. One might use another separator, but the idea is that one can never be sure which separator will definitely work before he actually checks the data.

Could be related with #17

[aneuway2]

Also seeing this issue in version 0.1.7 however degrading to version 0.1.6 does not show this issue