Attached is my solution. I used a Python tool to do the website scraping (because why not?). I navigate the pages if there are more than one (the provided scraping macro will miss data if the page count goes over 10) and I only add the valid solution types to the resulting data frame. I then get down to a unique record per user in order to not double count solution postings.
For the monies distribution, I went with quantity over quality (as the 4+ star criteria is already applied) as well as the largest variety of items. I also noticed that there were a few records with both the salesprice and msrp as null, I excluded those because I didn't figure anything was really free.
I thoroughly enjoyed this one. I went through @patrick_digan's idea of circling back with accumulated left over spend so I implemented that. I also split clothes up into boys and girls in order to make sure there was even spend there as well.
I also tried to make a pretty report but I wasn't able to get a report that actually rendered to match what I created in the "visual layout" tool in the laboratory.
Anyway. Here's Wonderwall
Here's my results:
I like how open-ended this challenge was!
By the way, there are not so many clothing items that qualify, so if we get more answers a workflow like this would benefit from making sure we do ten (at least a second) pass on the list of options.
Not to elegant but simple: Get 10 threads of the eligible items list, and "union" them into a large list before applying the "running totals".