create an option in core/merge_new_data to ignore invalid keys

[tinatn29]

Currently, merge_new_data(new_dataset) ignores keys that are in new_dataset but not in the existing self.dataset. We can make this optional and add those keys to self.dataset as new entries if a new argument 'ignore_invalid_keys=False'? + need to add more test cases for this

[sbillinge]

closed by #7