My original bug fix didn't work because I tested it for empty columns with None values it causes column dtype to be object and "infer_dtype" function interprets series dtype from its values. Upload with UI sets null values to np.nan, it causes data frame column dtype to be float and "infer_dtype" function interprets float series dtype from it's dtype not values.
I additionally fixed and added checks to test_return_error_message_when_not_correct_datatypes test function.
My original bug fix didn't work because I tested it for empty columns with None values it causes column dtype to be object and "infer_dtype" function interprets series dtype from its values. Upload with UI sets null values to np.nan, it causes data frame column dtype to be float and "infer_dtype" function interprets float series dtype from it's dtype not values. I additionally fixed and added checks to test_return_error_message_when_not_correct_datatypes test function.