[Apologies for wall of text! I didn't intend to do that, just got typing]
I don't think meaningful economic decisions are particularly fun gameplay.
I think the core game (sometimes), and newgrfs (frequently) reach for "it causes players to do cost-benefit" as a post-hoc rationalisation for "I had this idea that seemed good at the time". And because there are constant messages in the world that life is a cost-benefit, rational-maximisation exercise, we unthinkingly accept that this must also be a good model for good gameplay. But I find that doing economic cost-benefit is remarkably boring in a game.
But anyway, and so...
Trams have a different routing model to non-tram RVs, which adds variety to gameplay. Nothing in the many RoadType discussions I've seen (and there have been many) proposes additional routing models. Mostly the focus is on issues like appearance, catenary, sidewalks, speed limits, costs, modifying vehicle TE or power, and enabling/limiting vehicle types.
Road appearance is already modifiable via newgrf, and there are many road grfs. There is (limited) date support, offering dirt roads before a certain date.
Horse wagons are in eGRVTS, they're vanishingly unlikely to appear in core OpenTTD
Issues like overtaking and use of lanes are currently inherent limitations in OpenTTD's vehicle handling, and would not be touched by RoadTypes.
And...
To implement RoadTypes is a lot of work, given
- the limitations in the map array (there are free bytes, but only enough for a limited implementation)
- roads are inherently shared infrastructure, so any RoadTypes spec needs to be careful about accidental griefing by towns / other players / AIs. (Deliberate griefing is fine, if you don't like it, just don't play with idiots).
- multiple RoadTypes can be present on a tile. Different railtypes cannot. Handling this is non-trivial.
- roads need a relatively complex stack of sprites to handle cases like tram tracks and catenary.
- different types might contend to provide sprites like catenary.
There's no single blocking issue, but it adds up to a lot of fiddly work, and nobody is very interested so far in taking it on. Any debate about it typically ends in either "but first we need to rewrite the map array" or "what's the actual benefit to all this?".
And to end
Still nobody yet has made a compelling case for gameplay benefit of RoadTypes. But tbh, from what I've seen of development that's not how features tend to arrive anyway. Typically someone turns up with working code, and it gets in or it gets rejected. There is sometimes very careful up-front design and such (especially in newgrf spec and networking), and other times something looks fun and meets code standards so it goes in. I don't want to give a misleading impression that there is a constant rational debate about features, cos I'm not aware of one
So there's some more hot air expended by me on a feature that's never going to happen