As you alluded to in the original post, the most common approach is to use simulated switches to trigger scenes.
You could use the HTTP Action in SharpTools along with the Personal Access Token (like the one from the Scene Custom Tile). It’s probably slightly more efficient from an execution perspective, but I think it’s more complicated and a bit harder to wrap your head around.
I would also note that we have a next-generation integration with SmartThings in development which has native Scene support. We’re still working with SmartThings on ironing a few things out, but pushing hard on the remaining gaps with them as I’d really like to get a beta out. (Rather than use the Personal Access Token approach like is used with the Scene Custom Tile or with ActionTiles, we wanted to do a proper next-generation integration for the whole platform)