Some stuff I learned after the video for anyone who wants to try this:
A redditor offered to test my sample and found the case depth was only like .2.3 mm. I got the 30 minutes from a Clickspring video where he makes files, but it looks like he made the same mistake as me, assuming the visible difference in grain was the same as the case depth. If you're making a knife leave it in longer or you'll just sharpen away the high carbon layer. Double the time, double the depth and so on.
The normalizing cycles were unnecessary, as long as you did that after the forging there's no reason to do it after carburizing. All that will do is let the carbon diffuse more and burn off the surface. Best to quench in one heat, maybe even in water for something like this.