The thing that often gets either over-looked or not really discussed all that much is that 'frame building' is not a single skill.

Brazing, welding, filing/cutting mitres are all practical metalwork skills that can be (relatively) easily taught to engaged pupils and if they put the effort in they will perform them to an acceptable standard with enough practise. Certainly some will become more adept, some will have a 'natural flair' and learn techniques that allow them to braze/weld/file more consistently and reliably than others but the underlying skill is easily transferable and NOT magic, the real skill is in doing it consistently well.

However frame geometry is most definitely a learned skill and is as much mental as practical, although still has a very practical aspect to it. You can't become 'good' at frame geometry through repetition and practise alone, it requires you to study and comprehend the definable aspects, the maths and interactions between various forces and kinematics. But it also requires you to have direct experience of riding the damn things, or a VERY good and close relationship with someone who does who can feedback after riding your creations. Even if you arrive at a 'perfect' geometry for use case X under person Y though repetition and practical experience, that's not necessarily going to transfer to use case Z or another person.

So often I hear people saying things like "you don't want to put a a Xmm stem on that, it'll steer badly", or some other similar comment about a frame angle or whatnot, and yes some people have huge practical experience of riding many bikes, but more often than not this is 'repeated fact' rather than direct experience. How many people have deliberately ridden 'bad' combinations or setups to feel first hand how they ride? I've ridden many hundreds of bikes over the years, some far more than others, and can normally make a pretty good assessment of how any given bike will ride before I swing a leg over but there are still surprises, either from an aspect of the frame I wasn't aware of or how a particular component choice interacts with another etc. and I LOVE it when I get surprised. I also love to experiment too have been prompted to try various things after riding an unknown bike and it stirring up some thoughts about bits i liked/didn't like or found just plain different to what I'm used to.
Likewise, if you're building frames with a certain geometry I would hope that you've arrived at that decision based on thousands of hours and miles riding various geometries so that you really understand the effect, and if you're in the business of *custom* frames you need to be understanding the person riding it, the environment and use case they're riding in and building a frame to suit them, and very little of that will be defined by how good you are at brazing.

This also brings me on to tubing/component selection, which goes hand in hand with both geometry and the physical aspects of joining tubes together as well as how the user will interact with the frame (eg: fastenings, fittings, dropouts etc.). This is another area where expertise comes from practical experience as well as intellectual learning/hard maths. You can't go recommending a specific butting profile for a particular frame or rider if you don't actually know how that choice will affect the frame in terms of it's structural integrity and desired ride feel. You also need to have a decent understanding of the person who's going to pedal it, what their riding style is in terms of power delivery and output, how they sit on the bike, how they manhandle the bike at extremes and a dozen other factors.
This feeds back into your geometry choices too, changing one aspect can affect the other and this isn't knowledge gained from filing lugs day after day after day, it's gained from building, and riding (either directly or by proxy), large numbers of frames, in large numbers of situations, and not just good ones, you have to build bad ones too to understand why they're bad and others are good!

Working out how to join all those bits together the fulfill the requirements is a different skill to physically joining the components. Back in the olden days (I hate that phrase!) in bigger frame shops churning out large numbers of frames it would not be unusual for tasks to be distributed according to expertise, with the end frame being a mix of Johns design, Peters mitres, and Bobs brazing, and the frames were excellent and consistent.
To have one person able to perform all the tasks and building solo is not necessarily the best way of creating a frame, sure there are those that can do it, but it's not necessarily normal or optimal. In mass production we already have this, frames designed in one location, and built in another, but mass production throws in other issues like economy of scale, speed of production, and having to cater to a range of potential riders.
In custom building there's really no reason why you can't have a frame designed and components selected by one person, and physically joined by another, who is the frame builder? and who is responsible for how it rides?

I would much rather have a frame built to an average/acceptable physical standard by a builder who has the knowledge and experience to choose the right frame components and geometry than one who may be able to join each part exceptionally well but is joining a poor choice of tubes in a questionable arrangement!

TL;DR - anyone with the correct metalwork skills can 'build' a frame, they could even join all the tubes the the highest standard possible, but it won't necessarily be a 'good' frame.

That's where the skill is, in understanding and executing the whole.