No, I don't think you're crazy. If you have the inclination to experiment like that, I'd say go for it. If you do, we'd love to hear what you discover.
I'm thinking for the 2 special groups, just catch a set number of Beatles (10-20) from the main bin, put them in plastic shoe boxes for a set period of time (1-2 weeks), then put them back with the others. The ones for fastest life cycle, just take the first (10-20) pupae, set them to the side, and dump anything else back in the main bin, the pupae go back in the shoe box, once everyone is emerged and breeding, dump substrate again and give them fresh bedding for the same set period of time before putting them in with the others, that way the fastest will always be breeding with the fastest and the slower ones will be weeded out. Same basic thing with size, I would assume the pupae weigh the same (roughly) as the worm they came from, so taking the (10-20) largest pupae would be the selection criteria in that box.
Welcome to the thread! I don't think you're crazy at all, it sounds like a fun project. The first thing that comes to mind is that body size and rate of development are intricately intertwined. Generally speaking, it takes more time to achieve larger body sizes and, conversely, faster development tends to result in smaller size. I'm not sure if you can easily separate these two variables. It's worth trying though! And I think you have a decent start on the design of the project. A few thoughts come to mind about some issues you would encounter.
The animal behaviorist in me would want more detailed assessments of body size so that I would know whether the selection is working. Also, the variability in body size within a colony is so small that accurately choosing the largest individuals by simple visual inspection would be challenging, even more so after a few generations of selection when differences among individuals becomes smaller. Measuring body size in live beetles Is tough. Body weight would be an easier variable to measure, but you'd need a scale that measures in mg.
I would seek out several distinct sources of mealworms to start with so as to increase the genetic variability from which your selection can operate. I suspect that most beetles from a given source are highly inbred, which reduces genetic variability.
I'm sure more ideas will come to me later. Keep us updated on what you decide to do and bounce ideas off of us. We love to help.