Remove unnecessary dataframe copies in methods

[Erik-Geo]

In LayeredData.select_by_values, the copy operation is really slow on large datasets (e.g. DINO, 5 million rows):

selected = self.df.copy()

However, this also works as intended and doesn't slow down the select_by_values operation:

selected = self.df

See where we can optimize other methods by removing unnecessary copies and make sure that we test dataframe copy/view behaviour.