Lots of interesting comments here, but there are some gaps in logic.
First, 3-D printing. While it would reduce costs associated with tooling, other cost factors would remain high, and possibly skyrocket because of this approach. 3-D printing is fine for producing a shell, but a locomotive also includes hundreds of other parts, all of which would have to be designed to fit, interact, and integrate with each other to make a whole product. That means intensive design work. If you stray into territory that no one has in terms of locomotive production, you would have to add to the R&D costs. And you would be expected to pay the costs of amortization of the equipment needed to produce a model, 3-D printer included. Plus, we have a current alternative to minimal tooling options. It's called brass.
Second, making the model that you want. If it hasn't been made already, there is usually a reason why. While it's possible that manufacturers have missed one sure seller, it's much more likely that the model you want isn't one of them. Given the cost of traditional tooling, manufacturers have to spend a lot of time and effort deciding what to produce. Not surpringly, they have had very few misses. Market research and past experiences determine what will get manufactured. Or look at it this way. When you read someone else's idea of what should be made that hasn't been already, how excited are you about the prospect of spending $1,500 on it.
Third, the size of the market. This one is obvious. It's getting smaller, not larger. Producing a product in less demand than what has already been made becomes less likely. HO has the largest market, followed by N scale. By the time we get to O and S, the market becomes perceptibly finite. Think of it this way. Somewhere out there is someone saying, "Why doesn't someone make a Pennsy K4 in TT scale?" See what I mean?
I am not bashing people's desires. I have my list of wants too. I mean, where's my B&O C-16a? But, logic says we won't likely get our wishes fulfilled.
