In Azure Automated ML: how to retain original columns in the predict CSV output for a Test run, when they are not used in Input/ Output

Question

I have a Test dataset CSV with columns say X1, X2, X3, Y1. I use X2 and X3 as input and Y1 as output. When the prediction result comes it changes the column names
X2_orig, X3_orig and Y2_orig. since the column X1 is not used in Input and Output it is not even included.

Q1. How can I keep the original column names rather than appending "_orig" after each column name. I have to manually change them as I have post processing steps that use the original column names.

Q2. How can I keep the column X1 (the column unused as input/output) with all the row values in the result CSV? This column is a reference column and without this column I cannot do any post-processing.

Here is an example of the heading:

Y2_orig Y2_predicted D_predicted_proba U_predicted_proba X2_orig X3_orig X4_orig X6_orig X7_orig X8_orig X9_orig X10_orig X11_orig X12_orig X13_orig X14_orig X15_orig X16_orig X17_orig

These column does not includ the columns Y1, X5, Y1 which are reference columns and not used as input/ output.