If you've written anything in x86 assembly before and at least know how raycasting works 5 days actually seems like a pretty reasonable timeline. I imagine that a lot of that was trimming the raycast algorithm as much as possible. For instance, normally you do a fish-eye effect correction but that's clearly been skipped here.

As to how people come up with these ideas? It's kinda just like any hobby project, you just have an itch to try something, mull it over for a bit, and then decide to do it. Some people juggle geese...