I know this post is quite a few days old already, but I still wanted to add a bit to the discussion.

The printers you list vary wildly. Both in terms of design goals ("what is the printer meant to do well") and assembly requirement (from "ready to print in 10 minutes" to "you build this for like a week until something moves"). A Qidi is basically ready to go, a Prusa will take some time to put together (how much depends on if you got it as a kit or fully assembled). A Voron 2.4 takes about a week to build just for the printer, not including ERCF and/or tool changer, let alone tuning of said ERCF/tool changer.

Also there's the Troodon, which is a Formbot prebuilt that is closer to a real Voron 2.4 than a Sovol SV08, just to add to your list. It has a stock stealthburner tool head compared to the proprietary thing that Sovol uses, for example.

I've recently built a Voron 2.4r2 (Formbot kit) and loved it, but it was like my 4th printer (and a previous printer was a self-sourced scratch build). So do you have experience with 3d printers, and building them or tinkering with them? I would probably not recommend building one otherwise, but it's not impossible either, just expect a relatively steep learning curve if you have no prior experience.

Do you want to mostly just print in colors but same filament ype, or do you want to mainly have multi-material capabilities? So do you need 5+, or would 2 colors with the option to expand work for you?

If the Voron is a real option for you, I'd highly recommend it. Just make sure you're going with a can-bus based build/kit (like Formbot). These days I wouldn't go with an ERCF due to the complexity of building it and then setting it up, as tuning is supposedly a bit of a process. Also you mentioned that the amount of waste during multi-color prints is a real factor for you, and that puts single-nozzle systems inherently at a disadvantage as you just have to purge the hotend on every change. So I would suggest a tool changing system, and I would either start with that (but just 2 tool heads), or add it as the first project. Specifically, I would suggest using the Tapchanger as a modern system. Frankly adding a tool head like that is much less effort than building an ERCF, but also just adds 1 filament each and not like 9 at once.