Interesting ideas, but to play devils advocate, there are many problems with what you propose.
Primarily, this many SKU's is completely uneconomical for a company that's already seeing declining sales and profit margin. It's not just the number of models, it's the fact that they'll have to make multiple versions of each one for each country and carrier, and storage capacity.
-- HTC Universal: In addition to all the flavours of 3G/H+, you want support for all LTE frequencies? Good luck with that. Even assuming that it's technologically and financially feesible to cram that many different radios into one handset, it's still not useful. Many CDMA providers will not let you bring a phone to their network that has not been purchased through their stores. Even some GSM providers that can't block it will make it as difficult as they can. Even then, how many people really need access to more than 2 networks at most? The market would be incredibly small, and the cost of the phone would be enormous.
-- HTC Marathon: Interesting. But it's probably more reasonable to just sell one phone of any type with an option of multiple officially supported battery sizes.
-- HTC Pure: It's possible, and I'd buy it, but chances are it won't happen. Officially selling a non-Nexus pure-Android phone implies that your Sense brand is not as great as you'd like. So it's unlikely.
-- HTC Tinker: There is no way you'll ever get a phone that officially supports both Android and WP8. Microsoft would never allow that. And there is no convincing non-carrier reason you need to lock your bootloader on any device. Having a specific version just for the unlocked bootloader seems wasteful. Just unlock them all.
Overall, it makes more sense to just make one or two phones and include whatever of these options are feasible.
So instead of everything you proposed, they could just release the HTC One with an unlocked bootloader, varying internal storage, provide downloads for officially supported AOSP images, and multiple battery sizes. That's actually feasible and economical. That doesn't satisfy every possible niche, but it gets to the big ones, and the increase in production/engineering cost is much less significant.
But it still won't happen. Fact is that the cost of catering to these niches is probably far more than then the associated increase in revenue. Best you can hope for is an unlocked bootloader.