Hello,
Thanks to the detailed spec your requirements are quite clear and straightforward to implement. When you said we need to scroll down on the results page I thought the records were being fetched with xhr based on viewport position but it turned out that's a fake loading bar there. All the records are already fetched when the page is first requested and as the user scrolls down they're revealed with a simple css change so no problem in this regard.
There're a couple of things I want to mention about cron. As far as I can see there's a single output format YY_MM_DD-tor-spesa-media
which will be placed under "DATA" directory and 4 cron jobs you've suggested. The problem is if there's a chance that these would be run at the same time there'll be a single output file. Also, you don't need to specify 4 different cron jobs for each input file. The tool can check a designated folder, TORNEW in this case, to read the input file.
Anyway, per your request I'll use python to implement this. It'll be ready for you to test in 5 days at most, thanks.