From the company point of view that might make sense if they don't think about the consequences. But for the person, since we all know that usually any such projects are silently rejected without explanation it's totally not worth the effort.
Having been on the inside watching some people review take-home projects and arbitrarily reject them for the most random reasons ("I don't like the way he writes comments, what a loser") I also have zero faith on them being evaluated fairly.
The only way I'd ever consider doing any kind of endurance project for an interview is if the company guarantees results if I put in the effort. Could be something like: Write code to interact with and/or provide these well-documented APIs, it'll be run against a test suite where I can see the results and it needs to meet xyz performance criteria and if you can deliver it in X hours or less we guarantee a competitive offer. Only then would it be worth the work.
But to spend a lot of time just to be ghosted? Big no to that.