I assume you are referring to the part where article says that (hall effect sensor) "output being used to trigger MOSFETs". I have no idea where that came from, but in the schemeatics and code I looked at hall effect output was read using ADC like you would expect. Otherwise how else would you change the actuation point? You could digitise it by comparing it against adjustment pot with a discrete comparator IC, but that seems like 80% of hassle with 20% of functionality. Sample keypad in the linked repository doesn't do that. Either I am looking at different repository or Arstechnica writer blindly copied from the hackaday article, and hackaday writer was half asleep while reading schematic so wrote complete nonsense.

For a macro pad, maybe. For a full-sized keyboard, that would be a lot of ADC channels. Looking at Digikey, ADC chips seem to be about $0.50 per input in bulk, and that would add a lot of BOM cost to the keyboard.

You might be able to do better if you instead feed the Hall output into a sample-and-hold filter constructed out of a much cheaper op-amp. I suspect that would be an interesting engineering challenge.

Look at the Wooting keyboards. They do this. I'm sure there are others.