Nicely illustrated article, and stepped explanations. Easy to follow. It also demonstrates nomenclature and useage of script components. There are little trinkets of caveats that one must know to succesfully navigate the script component, and it appears you have worked it out.

This approach is an RBAR approach... even more so, not just row by row, but field by field in each row. As such would not be very flexible for change in column count.

Having dealt with this type of data many times in the education industry, there are often multiple course selections. In practice it has always served better, faster, and more flexible for future use/re-use of both the component and the data to normalize the data, no matter what the final output format may be. In a normalized format, it matters not if there are 1 or 100 CourseID columns, nor the size of the comparison set.